View
1
Download
0
Category
Preview:
Citation preview
Two-component spinor techniques and Feynman rules
for quantum field theory and supersymmetry
DRAFT version 1.15 April 4, 2008
Herbi K. Dreiner1, Howard E. Haber2 and Stephen P. Martin3
1Physikalisches Institut der Universitat Bonn, Nußallee 12, 53115 Bonn, Germany
2Santa Cruz Institute for Particle Physics, University of California, Santa Cruz CA 950643Department of Physics, Northern Illinois University, DeKalb IL 60115 and
Fermi National Accelerator Laboratory, P.O. Box 500, Batavia IL 60510
Abstract
We provide a complete set of Feynman rules for fermions using two-component spinornotation. These rules are suitable for practical calculations of cross-sections, decay rates, andradiative corrections in the Standard Model and its extensions, including supersymmetry.A unified treatment applies for massless Weyl fermions and massive Dirac and Majoranafermions. Numerous examples are given.
Contents
1 Introduction 3
2 Essential conventions and notations 6
3 Properties of fermion fields 14
3.1 A single two-component fermion field . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Fermion mass diagonalization and external wave functions in a general theory . . 20
4 Feynman rules with two-component spinors 24
4.1 External fermion rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2 Propagators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3 Fermion interactions with bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.4 General structure and rules for Feynman graphs . . . . . . . . . . . . . . . . . . 33
4.5 Basic examples of writing down diagrams and amplitudes . . . . . . . . . . . . . 34
4.6 Self-energy functions and pole masses for two-component fermions . . . . . . . . 42
5 Conventions for fermion and anti-fermion names and fields 49
6 Practical examples from the Standard Model and supersymmetry 54
6.1 Top quark decay: t→ bW+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2 Z0 vector boson decay: Z0 → f f . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
1
6.3 Bhabha scattering: e−e+ → e−e+ . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.4 Polarized Muon Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.5 Neutral Higgs boson decays φ0 → ff , for φ0 = h0,H0, A0 in supersymmetry . . . 61
6.6 Sneutrino decay νe → C+i e
− . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.7 Chargino Decay C+i → νee
+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.8 Neutralino Decays Ni → φ0Nj , for φ0 = h0,H0, A0 . . . . . . . . . . . . . . . . . 65
6.9 Ni → Z0Nj . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.10 Selectron pair production in electron-electron collisions . . . . . . . . . . . . . . . 67
6.10.1 e−e− → e−L e−R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.10.2 e−e− → e−Re−R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.10.3 e−e− → e−L e−L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.11 e−e+ → νν∗ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.12 e−e+ → NiNj . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.13 e−e+ → C−i C
+j . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.14 ud→ C+i Nj . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.15 Ni → NjNkN` . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.16 Three-body slepton decays ˜−R → `−τ± τ∓1 for ` = e, µ . . . . . . . . . . . . . . . 84
6.17 Neutralino decay to photon and Goldstino: Ni → γG . . . . . . . . . . . . . . . . 86
6.18 Gluino pair production from gluon fusion: gg → gg . . . . . . . . . . . . . . . . . 88
6.19 R-parity violating stau decay: τ+R → e+νµ . . . . . . . . . . . . . . . . . . . . . . 92
6.20 R-parity violating neutralino decay: Ni → µ−ud . . . . . . . . . . . . . . . . . . 94
6.21 Top-quark condensation from a Nambu-Jona-Lasinio model gap equation . . . . 96
6.22 Electroweak vector boson self-energies from fermion loops . . . . . . . . . . . . . 98
6.23 Self-energy and pole mass of the top quark . . . . . . . . . . . . . . . . . . . . . 101
6.24 Self-energy and pole mass of the gluino . . . . . . . . . . . . . . . . . . . . . . . . 108
6.25 Triangle anomaly from chiral fermion loops . . . . . . . . . . . . . . . . . . . . . 111
Appendix A: Two-component spinor identities in d 6= 4 115
Appendix B: Explicit forms for the two-component spinor wave functions 117
Appendix C: Path integral treatment of two-component fermion propagators 121
Appendix D: Matrix decompositions for fermion mass diagonalization 126
D.1 Singular Value Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
D.2 Takagi Diagonalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
D.3 Relation between Takagi diagonalization and the singular value decomposition . . 130
2
Appendix E: Correspondence to four-component spinor notation 132
E.1 Dirac matrices and four-component spinors . . . . . . . . . . . . . . . . . . . . . . 132
E.2 Feynman rules for four-component fermions . . . . . . . . . . . . . . . . . . . . . . 137
E.3 Applications of four-component spinor Feynman rules . . . . . . . . . . . . . . . . 142
E.4 Self-energy functions and pole masses for four-component fermions . . . . . . . . . 147
Appendix F: Covariant spin operators and the Bouchiat-Michel Formulae 149
F.1 The covariant spin operators for a spin-1/2 fermion . . . . . . . . . . . . . . . . . 149
F.2 Two-component spinor wave function relations . . . . . . . . . . . . . . . . . . . . 150
F.3 Two-component Bouchiat-Michel formulae . . . . . . . . . . . . . . . . . . . . . . 152
F.4 Four-component Bouchiat-Michel formulae . . . . . . . . . . . . . . . . . . . . . . 155
Appendix G: The helicity amplitude techinique 157
Appendix H: Standard Model fermion interaction vertices 159
Appendix I: MSSM Fermion Interaction Vertices 166
I.1 Higgs-fermion interaction vertices in the MSSM . . . . . . . . . . . . . . . . . . . . 166
I.2 Gauge interaction vertices for neutralinos and charginos . . . . . . . . . . . . . . . 170
I.3 Higgs interactions with charginos and neutralinos . . . . . . . . . . . . . . . . . . . 172
I.4 Chargino and neutralino interactions with fermions and sfermions . . . . . . . . . . 173
I.5 SUSY QCD Feynman Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Appendix J: Trilinear R-Parity Violating Fermion Interaction Vertices 181
1 Introduction
A crucial feature of the Standard Model of particle physics is the chiral nature of fermion
quantum numbers and interactions. According to the modern understanding of the electroweak
symmetry, the fundamental degrees of freedom for quarks and leptons are two-component Weyl-
van der Waerden fermions, i.e. 2-component spinors under the Lorentz group, that transform
as irreducible representations under the gauge group SU(2)L × U(1)Y . Furthermore, within
the context of supersymmetric field theories, two-component spinors enter naturally, due to the
spinor-nature of the symmetry generators themselves, as well as through the holomorphic nature
of the superpotential. Despite this, most pedagogical treatments and practical calculations in
high-energy physics continue to use the four-component Dirac notation, which combines distinct
irreducible representations of the symmetry groups. Parity-conserving theories such as QED and
QCD are well-suited to the four-component fermion methods. There is also a certain perceived
advantage to familiarity. However, as we progress to phenomena at and above the scale of
3
electroweak symmetry breaking, it seems increasingly natural to employ two-component fermion
notation, in harmony with the irreducible transformation properties dictated by the physics.
One occasionally encounters the misconception that two-component fermion notations are
somehow inherently ill-suited or unwieldy for practical use. Perhaps this is due in part to a
lack of examples of calculations using two-component language in the pedagogical literature.
In this paper, we seek to dispel this idea by presenting Feynman rules for fermions using two-
component spinor notation, intended for practical calculations of cross-sections, decays, and
radiative corrections. This formalism employs a unified framework that applies equally well to
Dirac fermions like the Standard Model quarks and leptons, and to Majorana fermions such as
the light neutrinos that appear in the seesaw extension of the Standard Model1 or the neutralinos
of the minimal supersymmetric extension of the Standard Model (MSSM) [47–49].
Spinors were introduced by E. Cartan in 1913 as projective representations of the rota-
tion group [1]. They entered into physics via the Dirac equation in 1928 [3]. In the same year,
H. Weyl discussed the representations of the Lorentz group, including the two-component spinor
representations, in terms of stereographic projective coordinates [4]. The extension of the tensor
calculus (or tensor analysis) to spinor calculus (spinor analysis) was given by B. L. van der Waer-
den [5], upon instigation of P. Ehrenfest. It is in this paper also that v. d. Waerden (not Weyl
as often claimed in the literature) first introduces the notation of dotted and undotted indices
for the irreducible ( 12 ,0) and (0, 12 ) representations of the Lorentz group. Both Weyl and van
der Waerden independently consider the decomposition of the Dirac equation into two coupled
differential equations for two-component spinors. Early, more pedagogical discussions of two-
component spinors are given in [6–8]. See also [9]. Ref. [6], is also the first English paper to
employ the dotted and undotted index notation. Very nice early reviews on spinor techniques
were written by Bade and Jehle in 1953 [10] and in German by Cap in 1954 [11].
Two component spinors have also been discussed in many non-supersymmetric textbooks,
see for example [4,12–31]. Among the early books, we would like to draw attention to [12], which
has an extensive discussion and the appendix of [19]. Scheck [21] includes a short discussion of
the field theory of two-component spinors, including the propagator. The most extensive field
theoretic discussion is given by Ticciati [24]. This includes a complete set of Feynman rules for
a Yukawa theory as well as three example calculations. recently, Srednicki [31] has written an
introduction into quantum field theory with a dsicussion of two-component fermions, including
their quantization.
[I have not checked the books [18, 26–28, 30] since they are not in our library.]
All text books on supersymmetry [32–45] include a discussion of two-component spinors
on some level. This also typically includes a discussion of dotted and undotted indices as well
1In the limit of zero mass, neutrinos can be described by either Majorana or Weyl fermions. Both are naturallydescribed in the two-component fermion formalism.
4
as a collection of identities involving the sigma matrices. Particularly extensive and useful sets
of identities can be found in [32, 37, 38, 42, 43]. Terning [43] also includes some field theoretic
details.
[This part is about the previous work on van der Waerden spinors in particle physics pa-
pers.] The standard technique for computing scattering cross sections with initial and final
state fermions involves squaring the amplitudes, summing over the spin states and then com-
puting the traces of products of gamma matrices, or in the two-component case, over products
of sigma matrices. We employ this latter technique throughout this paper. However, the com-
putational effort rises with the square of the number of interfering diagrams. This typically
becomes impractical with four or more particles in the final state. One approach to make such
extensive calculations manageable, is the helicity amplitude technique. Here the scattering pro-
cess is decomposed into the scattering of helicity eigenstates. Then the individual amplitudes
are computed analytically in terms of Lorentz scalar invariants, i.e. a complex number, which
can be readily computed. It is then a simple numerical task to sum the amplitudes and square
them. This was first explored in refs [51–54], using four-component spinors, see also refs [55–59].
For spinor techniques in th helicity formalism see also [60]. The natural spinor formalism for
the helicity amplitude techniques are in fact the 2-component Weyl-van der Waerden spinors,
which we discuss in detail in this paper. They were implemented in the helicity amplitude
technique in refs. [61–66]. Recently the two-component formalism has been implemented in a
computer program for the numerical computation of amplitudes and cross sections for event
generators multi-particle processes refs [133, 134]. In order to to see how to apply our work to
the case of multi-particle final states, we present in Appendix G the translation between our
notation and that of the widely used Hagiwara and Zeppenfeld (HZ) formalism ref. [63]. It is
then straightforward to implement amplitudes as computed here into a numerical cross section
computation.
[Something like this is still missing, I just sketched it.] This report is outlined as follows.
In Sect. 2, we present our conventions and notation. In Sect. 3 we derive the basic properties
of the quantized two-component fermion fields. In Sect. 4 we derive the Feynman rules for
two-component spinors and describe how to write down amplitudes in our formalism. In Sect. 5
we give our convention for fermion and anti-fermion names and fields. This is important for
consistently writing down the amplitudes for a given physical process and also for comparing
with previous 4-component computations. In Sect. 6 we then compute an extensive number of
examples using our formalism. This is the central part of our paper.
5
2 Essential conventions and notations
We begin with a discussion of necessary conventions. The metric tensor is taken2 to be:
gµν = diag(+1,−1,−1,−1) , (2.1)
where µ, ν, ρ . . . = 0, 1, 2, 3 are spacetime vector indices. Contravariant four-vectors (e.g. posi-
tions and momenta) are defined with indices raised, and covariant four-vectors (e.g. derivatives)
with lowered indices:
xµ = (t , ~x), (2.2)
pµ = (E , ~p), (2.3)
∂µ ≡∂
∂xµ= (∂/∂t , ~∇) , (2.4)
in units with c = 1. The totally antisymmetric pseudo-tensor εµνρσ is defined such that
ε0123 = −ε0123 = +1 . (2.5)
The irreducible building blocks for spin-1/2 fermions are fields that transform either under
the left-handed ( 12 , 0) or the right-handed (0, 1
2 ) representation of the Lorentz group. Hermitian
conjugation interchanges these two representations. A massive Majorana fermion field can be
constructed from either representation; this is the spin-1/2 analog of a real scalar field. The
Dirac field combines two equal mass two-component fields into a reducible representation of the
form (12 , 0) ⊕ (0, 1
2); this is the spin-1/2 analog of a complex scalar field. It is also possible to
use four-component notation to describe Majorana fermions by imposing a reality condition on
the spinor in order to reduce the number of degrees of freedom in half. However, in this paper,
we shall focus primarily on two-component spinor notation for all fermions. In the following,
(12 , 0) spinors carry undotted indices α, β, . . . = 1, 2, and (0, 1
2) spinors carry dotted indices
α, β, . . . = 1, 2.
We begin by briefly considering the representations of the Lorentz group. Under a Lorentz
transformation, a contravariant four-vector xµ transforms as
xµ → x′µ = Λµνxν , (2.6)
2The published version of this paper uses the (+,−,−,−) metric. An otherwise identical version, using the(−, +, +,+) metric favored by one of the authors, may be found at http://zippy.physics.niu.edu/rules.html.It can also be constructed by changing a single macro within the LATEX source file, in an obvious way. You cantell which version you are presently reading from equation (2.1). In general, the relative minus sign needed toswitch between one metric signature and the other is given by:
(−1)(Nσ+Nm+Nd)
where Nσ is the total number of σ and σ matrices, Nm is the number of metric tensors appearing either explicitlyor implicitly through contracted upper and lower indices, and Nd is the number of spacetime derivatives. Thisapplies to any relativistically covariant term appearing additively in a valid equation.
6
where Λ satisfies ΛµνgµρΛρλ = gλν . It then follows that the transformation of the corresponding
covariant four-vector xµ ≡ gµνxµ satisfies:
xν = x′µΛµν . (2.7)
The most general proper orthochronous Lorentz transformation (which is continuously connected
to the identity), corresponding to a rotation by an angle θ about an axis n [~θ ≡ θn] and a boost
vector ~ζ ≡ v tanh−1 β [where v ≡ ~v/|~v| and β ≡ |~v|], is a 4× 4 matrix given by:
Λ = exp(− i
2θρσsρσ
)= exp
(−i~θ ·~s− i~ζ ·~k
), (2.8)
where θi ≡ 12εijkθjk, ζ
i ≡ θi0 = −θ0i, si ≡ 12εijksjk, k
i ≡ s0i = −si0 and
(sρσ)µν = i(gρ
µ gσν − gσµ gρν) . (2.9)
Here, the indices i, j, k = 1, 2, 3 and ε123 = +1.
It follows from eqs. (2.8) and (2.9) that an infinitesimal orthochronous Lorentz transforma-
tion is given by Λµν ' δµν + θµν (after noting that θµν = −θνµ). Moreover, the infinitesimal
boost parameter is ~ζ ≡ v tanh−1 β ' βv ≡ ~β, since β 1 for an infinitesimal boost. Hence,
the actions of the infinitesimal boosts and rotations on the spacetime coordinates are
Rotations:
~x→ ~x′ ' ~x + (~θ × ~x)
t→ t′ ' t(2.10)
Boosts:
~x→ ~x′ ' ~x + ~βt
t→ t′ ' t+ ~β·~x ,(2.11)
with exactly analogous transformations for any contravariant four-vector.
With respect to the Lorentz transformation Λ, a general n-component field Φ transforms
as Φ(xµ)→ MR(Λ)Φ′(x′µ), where MR(Λ) is a (finite) n-dimensional matrix representation, R,
of the Lorentz group. Equivalently, the functional form of the transformed field Φ obeys
Φ′(xµ) = MR(Λ)Φ([Λ−1]µνxν). (2.12)
For proper orthochronous Lorentz transformations,
MR = exp
(− i
2θµνJ
µν
)' I − i~θ · ~J − i~ζ · ~K , (2.13)
where θµν parameterizes the Lorentz transformation Λ [eq. (2.8)], and J µν is a matrix-valued
antisymmetric tensor corresponding to the representation R. For infinitesimal Lorentz trans-
formations, we identify ~J and ~K as the generators of rotations parameterized by ~θ and boosts
parameterized by ~ζ, respectively. These three-vector generators are related to J µν by
J i ≡ 12εijkJjk , Ki ≡ J0i . (2.14)
7
Here we focus on the simplest non-trivial irreducible representations of the Lorentz algebra.
These are the two-dimensional (inequivalent) representations: ( 12 , 0) and (0, 1
2). In the ( 12 , 0)
representation, ~J = ~σ/2 and ~K = −i~σ/2 in eq. (2.13), where ~σ are the Pauli matrices. This
yields
M(12 ,0)≡M ' I − i~θ ·~σ/2− ~ζ ·~σ/2 . (2.15)
By definition M carries undotted spinor indices, as indicated by Mαβ. A two-component ( 1
2 , 0)
spinor is denoted by ψα and transforms as ψα → Mαβψβ, omitting the coordinate arguments
of the fields, which are as in eq. (2.12). Note that in our conventions for the location of the
spinor indices, one sums over a repeated index pair in which one index is lowered and one index
is raised.
In the (0, 12) representation, ~J = −~σ∗/2 and ~K = −i~σ∗/2 in eq. (2.13), so that its repre-
sentation matrix is M ∗, the complex conjugate of eq. (2.15). By definition, the indices carried
by M∗ are dotted, as indicated by (M ∗)αβ . A two-component (0, 12) spinor is denoted by ψα and
transforms as ψα → (M∗)αβψβ , again suppressing the coordinate arguments of the fields, which
are as in eq. (2.12). The reason for distinguishing between the undotted and dotted spinor index
types is that they cannot be directly contracted with each other to form a Lorentz invariant
quantity.
It follows that the (0, 12 ) and (1
2 , 0) representations can be related by complex conjugation.
That is, if ψα is a (0, 12 ) fermion, then (ψα)∗ transforms as a ( 1
2 , 0) fermion. This means that
we can, and will, describe all fermion degrees of freedom using only fields defined as left-handed
(12 , 0) fermions ψα, and their conjugates. We can combine spinors to make Lorentz tensors, so
it is useful to regard ψα as a row vector, and ψα as a column vector, with:
ψα ≡ (ψα)†. (2.16)
A check of the Lorentz transformation property of ψα then follows from (ψα)† → (ψβ)†(M †)β α,
where (M †)β α = (M∗)αβ reflects the definition of the hermitian adjoint matrix as the complex
conjugate transpose of the matrix. Again the coordinate arguments of the fields have been
suppressed, and are as in eq. (2.12). We will use the dotted-index notation in association
with the bar over the symbol as a synonym for hermitian conjugation, as above. [Many other
references write ψ†α to mean the same thing as eq. (2.16).]
There are two additional spin-1/2 irreducible representations of the Lorentz group, (M −1)T
and (M−1)†, but these are equivalent representations to the ( 12 , 0) and the (0, 1
2) representations,
respectively. The spinors that transform under these representations have raised spinor indices,
e.g., ψα and ψα, respectively. The spinor indices are raised and lowered with the two-index
antisymmetric symbol with components ε12 = −ε21 = ε21 = −ε12 = 1, and the same set of sign
conventions for the corresponding dotted spinor indices. Thus
ψα = εαβψβ , ψα = εαβψβ , ψα = εαβψ
β , ψα = εαβψβ . (2.17)
8
The ε symbol satisfies:3
εαβεγδ = −δγαδδβ + δδαδ
γβ , εαβε
γδ = −δγαδδβ + δδαδγ
β, (2.18)
from which it follows that:
εαβεβγ = εγβεβα = δγα, εαβε
βγ = εγβεβα = δγα . (2.19)
εαβ (εαβ) is also called the “spinor metric tensor” since it raises (lowers) spinor indices. It was
first introduced in this context in [5], but see also [6, 7, 10, 50] for related early work.
To construct Lorentz invariant Lagrangians and observables, one needs to first combine
products of spinors to make objects that transform as Lorentz tensors. In particular, Lorentz
vectors are obtained by introducing the sigma matrices σµαβ and σαβµ defined by [4, 5, 7, 8]
σ0 = σ0 =
(1 0
0 1
), σ1 = −σ1 =
(0 1
1 0
),
σ2 = −σ2 =
(0 −ii 0
), σ3 = −σ3 =
(1 0
0 −1
). (2.20)
The σ-matrices above have been defined with a lower (covariant) index. We also define the
corresponding quantities with upper (contravariant) indices:
σµ = gµνσν = (I2 ; ~σ) , σµ = gµνσν = (I2 ; −~σ) , (2.21)
where I2 is the 2× 2 identity matrix. The relations between σµ and σµ are
σµαα = εαβεαβσµ ββ , σµ αα = εαβεαβσµ
ββ, (2.22)
εαβσµβα = εαβσµβα , εαβσµ
αβ= εαβσ
µαβ . (2.23)
In general, just like tensors, we can have spinor objects with more than one spinor index:
Sα1α2...αnβ1β2...βn, where each α-index transforms separately according to Mα′
i
αi in eq. (2.15)
and each β-index transforms according to (M ∗)β′i
βi . Using the above σβαµ there is a one-to-one
correspondence between each bi-spinor Vαβ and a corresponding Lorentz four-vector V µ
V µ ≡ σβαµ Vαβ (2.24)
3Various subsets of the subsequent identities in this section involving commuting and non-commuting two-component spinors, as well as the ε and σ-matrices appear in many books, and papers, e.g. the books [8, 32–45],and the papers [61–66]
9
When constructing Lorentz tensors from fermion fields, the heights of spinor indices must
be consistent in the sense that lowered indices must only be contracted with raised indices. As
a convention, indices contracted like
αα and α
α , (2.25)
can be suppressed. In all spinor products given in this paper, contracted indices always have
heights that conform to eq. (2.25). For example,
ξη ≡ ξαηα, (2.26)
ξη ≡ ξαηα, (2.27)
ξσµη ≡ ξασµαβηβ , (2.28)
ξσµη ≡ ξασµαβηβ. (2.29)
As previously noted, it is convenient to regard ηα as a column vector and ξα as a row vector.
Consequently, if we also regard ξα as a row vector and ηα as a column vector then all the
spinor-index contraced products above have natural interpretations as products of matrices and
vectors.
The behavior of the spinor products under hermitian conjugation (for quantum field oper-
ators) or complex conjugation (for classical fields) is as follows:
(ξη)† = ηξ, (2.30)
(ξσµη)† = ησµξ, (2.31)
(ξσµη)† = ησµξ, (2.32)
(ξσµσνη)† = ησνσµξ . (2.33)
More generally,
(ξΣη)† = ηΣrξ , (2.34)
(ξΣη)† = ηΣr ξ , (2.35)
where in each case Σ stands for any sequence of alternating σ and σ matrices, and Σr is obtained
from Σ by reversing the order of all of the σ and σ matrices. Note that eqs. (2.30)–(2.35) are
applicable both to anti-commuting and to commuting spinors.
The following identities can be used to systematically simplify expressions involving prod-
10
ucts of σ and σ matrices:
σµαασββµ = 2δβαδ
βα , (2.36)
σµαασµββ = 2εαβεαβ , (2.37)
σµαασββµ = 2εαβεαβ , (2.38)
[σµσν + σνσµ]αβ = 2gµνδβα , (2.39)
[σµσν + σνσµ]αβ = 2gµνδαβ, (2.40)
σµσνσρ = gµνσρ − gµρσν + gνρσµ + iεµνρκσκ , (2.41)
σµσνσρ = gµνσρ − gµρσν + gνρσµ − iεµνρκσκ . (2.42)
Computations of cross sections and decay rates generally require traces of alternating products
of σ and σ matrices (see for example [62]):
Tr[σµσν ] = Tr[σµσν ] = 2gµν , (2.43)
Tr[σµσνσρσκ] = 2 (gµνgρκ − gµρgνκ + gµκgνρ + iεµνρκ) , (2.44)
Tr[σµσνσρσκ] = 2 (gµνgρκ − gµρgνκ + gµκgνρ − iεµνρκ) . (2.45)
Traces involving a larger even number of σ and σ matrices can be systematically obtained from
eqs. (2.43)–(2.45) by repeated use of eqs. (2.39) and (2.40) and the cyclic property of the trace.
Traces involving an odd number of σ and σ matrices cannot arise, since there is no way to
connect the spinor indices consistently.
In addition to manipulating expressions containing anticommuting fermion fields, we of-
ten must deal with products of commuting spinor wave functions that arise when evaluating
the Feynman rules. In the following expressions we denote the generic spinor by zi. In the
various identities listed below, an extra minus sign arises when manipulating a product of anti-
commuting fermion fields. Thus, we employ the notation:
(−1)A ≡
+1 , commuting spinors,
−1 , anticommuting spinors.(2.46)
The following identities hold for the zi:
z1z2 = −(−1)Az2z1 (2.47)
z1z2 = −(−1)Az2z1 (2.48)
z1σµz2 = (−1)Az2σ
µz1 (2.49)
z1σµσνz2 = −(−1)Az2σ
νσµz1 (2.50)
z1σµσν z2 = −(−1)Az2σ
νσµz1 (2.51)
z1σµσρσνz2 = (−1)Az2σ
νσρσµz1 , (2.52)
11
and so on.
Two-component spinor products can often be simplified by using Fierz identities. The
antisymmetry of the suppressed two-component ε symbol [eq. (2.18)] implies the identities:
(z1z2)(z3z4) = −(z1z3)(z4z2)− (z1z4)(z2z3) , (2.53)
(z1z2)(z3z4) = −(z1z3)(z4z2)− (z1z4)(z2z3) , (2.54)
where we have used eqs. (2.47) and (2.48) to cancel out any residual factors of (−1)A. Similarly,
eq. (2.36) can be used to derive
(z1σµz2)(z3σµz4) = −2(z1z4)(z2z3) , (2.55)
(z1σµz2)(z3σµz4) = 2(z1z3)(z4z2) , (2.56)
(z1σµz2)(z3σµz4) = 2(z1z3)(z4z2) . (2.57)
Eqs. (2.53)–(2.57) hold for both commuting and anticommuting spinors. Other Fierz identities
for spinors can be constructed trivially from these by appropriate choices of z1, z2, z3, and z4.
From the sigma matrices, one can construct the antisymmetrized products:
(σµν)αβ ≡ i
4
(σµαγσ
νγβ − σναγσµγβ), (2.58)
(σµν)αβ ≡i
4
(σµαγσνγβ − σναγσµγβ
). (2.59)
The matrices σµν and σµν satisfy self-duality relations
σµν = −12 iε
µνρκσρκ , σµν = 12 iε
µνρκσρκ . (2.60)
In addition, eq. (2.18) implies that
εβρεατσµνα
β = σµνρτ , εβρεατσ
µναβ = σµν ρτ , (2.61)
εατσµναβ = ερβσµνρ
τ , εατσµνα
β = ερβσµν ρ
τ , (2.62)
where we have used Tr(σµν) = Tr(σµν) = 0.
The σµν and σµν can be identified as the generators Jµν [see eq. (2.13)] of the Lorentz
group in the ( 12 , 0) and (0, 1
2 ) representations, respectively. That is, for the ( 12 , 0) representation
with a lowered undotted index (e.g. ψα), Jµν = σµν , while for the (0, 1
2) representation with a
raised dotted index (e.g. ψα), Jµν = σµν . In particular, the infinitesimal forms for the 4 × 4
Lorentz transformation matrix Λ and the corresponding matrices M and (M−1)† that transform
the (12 , 0) and (0, 1
2) spinors, respectively, are given by:
M ' I2 −i
2θµνσ
µν , (2.63)
(M−1)† ' I2 −i
2θµνσ
µν , (2.64)
Λµν ' δµν + 12
(θανg
αµ − θνβgβµ). (2.65)
12
The inverses of these quantities are obtained (to first order in θ) by replacing θ → −θ in the
above formulae. Using these infinitesimal forms [with the assistance of eqs. (A.18)–(A.21)], one
can establish the following two results:
M †σµM = Λµν σν , (2.66)
M−1σµ(M−1)† = Λµν σν . (2.67)
Eqs. (2.66) and (2.67) can be used to prove the covariance properties (with respect to Lorentz
transformations) of the transformation law for the two-component undotted and dotted spinor
fields, respectively.
As an example, consider a pure boost from the rest frame to a frame where pµ = (Ep , ~p),
which corresponds to θij = 0 and ζ i = θi0 = −θ0i. The matrices Mαβ and [(M−1)†]αβ that
govern the Lorentz transformations of spinor fields with a lowered undotted index and spinor
fields with a raised dotted index, respectively, are given by:
exp
(− i
2θµνJ
µν
)=
M = exp(−1
2~ζ · ~σ
)=
√p·σm
, for (12 , 0) ,
(M−1)† = exp(
12~ζ · ~σ
)=
√p·σm
, for (0, 12 ) ,
(2.68)
where4
√p·σ ≡ Ep +m− ~σ ·~p√
2(Ep +m), (2.69)
√p·σ ≡ Ep +m+ ~σ ·~p√
2(Ep +m). (2.70)
These matrix square roots are defined under the assumption that p0 = E~p ≡ (|~p|2 +m2)1/2, and
are chosen to be the unique hermitian matrices with non-negative eigenvalues whose squares are
equal to p·σ and p·σ, respectively.
Consider an arbitrary four-vector Sµ [defined in a referecne frame where pµ = (E ; ~p)],
whose rest frame value is SµR, i.e.
Sµ = ΛµνSνR , with Λ =
E/m pj/m
pi/m δij +pipj
m(E +m)
. (2.71)
Then, using eqs. (2.7), (2.67) and (2.68), it follows that:
√p·σ S ·σ√p·σ = mSR ·σ , (2.72)
√p·σ S ·σ
√p·σ = mSR ·σ . (2.73)
4One can check the validity of eqs. (2.69) and (2.70) by squaring both sides of the respective equations.
13
The generalization of the spinor results of this section to d 6= 4, useful for dimensional con-
tinuation regularization schemes, is discussed in Appendix A. In particular, the Fierz identities
of eqs. (2.36)–(2.37) and eqs. (2.55)–(2.57) and the identities (2.41), (2.42), (2.44) and (2.45)
involving the 4-dimensional ε tensor are not valid unless µ is a Lorentz vector index in exactly
4 dimensions. In d 6= 4 dimensions, as used for loop amplitudes in dimensional regularization
and dimensional reduction schemes, the necessary modifications are given in Appendix A. We
also direct the reader’s attention to Appendix E, which gives a detailed correspondence between
two-component spinor and four-component spinor notations.
3 Properties of fermion fields
3.1 A single two-component fermion field
We begin by describing the properties of a free neutral massive anti-commuting spin-1/2 field,
denoted ξα(x), which transforms as ( 12 , 0) under the Lorentz group. The field ξα therefore
describes a Majorana fermion. The free-field Lagrangian density is [5–7]:
L = iξσµ∂µξ − 12m(ξξ + ξξ) . (3.1)
On-shell, ξ satisfies the free-field Dirac equation [4, 5],
iσµαβ∂µξβ = mξα . (3.2)
Consequently after quantization, ξα can be expanded in a Fourier series [46]:
ξα(x) =∑
s
∫d3~p
(2π)3/2(2Ep)1/2
[xα(~p, s)a(~p, s)e
−ip·x + yα(~p, s)a†(~p, s)eip·x
], (3.3)
where Ep ≡ (|~p|2 + m2)1/2, and the creation and annihilation operators a† and a satisfy anti-
commutation relations:
a(~p, s), a†(~p ′, s′) = δ3(~p− ~p ′)δss′ , (3.4)
and all other anticommutators vanish. It follows that
ξα(x) ≡ (ξα)† =
∑
s
∫d3~p
(2π)3/2(2Ep)1/2
[xα(~p, s)a†(~p, s)eip·x + yα(~p, s)a(~p, s)e
−ip·x]. (3.5)
We employ covariant normalization of the one particle states, i.e., we act with one creation
operator on the vacuum with the following convention
|~p, s〉 ≡ (2π)3/2(2Ep)1/2a†(~p, s) |0〉 , (3.6)
so that⟨~p, s|~p ′, s′
⟩= (2π)3(2Ep)δ3(~p− ~p ′)δss′ . Therefore,
〈0| ξα(x) |~p, s〉 = xα(~p, s)e−ip·x , 〈0| ξα(x) |~p, s〉 = yα(~p, s)e
−ip·x , (3.7)
〈~p, s| ξα(x) |0〉 = yα(~p, s)eip·x , 〈~p, s| ξα(x) |0〉 = xα(~p, s)eip·x . (3.8)
14
It should be emphasized that ξα(x) is an anticommuting spinor field, whereas xα and yα are
commuting two-component spinor wave functions. The anticommuting properties of the fields
are carried by the creation and annihilation operators.
Applying eq. (3.2) to eq. (3.3), we find that the xα and yα satisfy momentum space Dirac
equations. These conditions can be written down in a number of equivalent ways:
(p·σ)αβxβ = myα , (p·σ)αβ yβ = mxα , (3.9)
(p·σ)αβ xβ = −myα , (p·σ)αβyβ = −mxα , (3.10)
xα(p·σ)αβ = −myβ , yα(p·σ)αβ = −mxβ , (3.11)
xα(p·σ)αβ = myβ , yα(p·σ)αβ = mxβ . (3.12)
Using the identities [(p·σ)(p·σ)]αβ = p2 δα
β and [(p·σ)(p·σ)]αβ = p2 δαβ, one can quickly check
that both xα and yα must satisfy the mass-shell condition, p2 = m2 (or equivalently, p0 = Ep).
We will later see that eqs. (3.9)–(3.12) are often useful for simplifying matrix elements.
The quantum number s labels the spin or helicity of the spin-1/2 fermion. We shall consider
two approaches for constructing the spin-1/2 states. In the first approach, we consider the
particle in its rest frame and quantize the spin along a fixed axis specified by the unit vector
s ≡ (sin θ cosφ , sin θ sinφ , cos θ) with polar angle θ and azimuthal angle φ with respect to a
fixed z-axis.5 The corresponding spin states will be called fixed-axis spin states. The relevant
basis of two-component spinors χs are eigenstates of 12~σ ·s, i.e.,
12~σ ·sχs = sχs , s = ±1
2 . (3.13)
Explicit forms for the two-component spinors χs and their properties are given in Appendix B.
The fixed-axis spin states described above are not very convenient for particles in relativistic
motion. Moreover, these states cannot be empoyed for massless particles since no rest frame
exists. Thus, a second approach is to consider helicity states and the corresponding basis of
two-component helicity spinors χλ that are eigenstates of 12~σ ·p, i.e.,
12~σ ·pχλ = λχλ, λ = ±1
2 . (3.14)
Here p is the unit vector in the direction of the three-momentum, with polar angle θ and
azimuthal angle φ with respect to a fixed z-axis. That is, the two-component helicity spinors
can be obtained from the fixed-axis spinors by replacing s by p and identifying θ and φ as the
polar and azimuthal angles of p.
For fermions of mass m 6= 0, it is possible to define the spin four-vector Sµ, which is specified
in the rest frame by (0; s). The unit three-vector s corresponds to the axis of spin quantization
5In the literature, it is a common practice to choose s = z. However in order to be somewhat more general,we shall not assume this convention here.
15
in the case of fixed-axis spin states. In an arbitrary reference frame, the spin four-vector satisfies
S ·p = 0 and S ·S = −1. After boosting from the rest frame to a frame in which pµ = (E , ~p)
[cf. eq. (2.71)], one finds:
Sµ =
(~p·s
m; s +
(~p·s) ~p
m(E +m)
). (3.15)
If necessary, we shall write Sµ(s) to emphasize the dependence of Sµ on s.
The spin four-vector for helicity states is defined by taking s = p. Eq. (3.15) then reduces to
Sµ =1
m(|~p| ; Ep) . (3.16)
In the non-relativistic limit, the spin four-vector for helicity states is Sµ ≈ (0 ; p), as expected.6
In the high energy limit (E m), Sµ = pµ/m + O(m/E). For a massless fermion, the spin
four-vector does not exist (as there is no rest frame). Nevertheless, one can obtain consistent
results by working with massive helicity states and taking the m → 0 limit at the end of the
computation. In this case, one can simply use Sµ = pµ/m+O(m/E); in practical computations
the final result will be well-defined in the zero mass limit. In contrast, for massive fermions at
rest, the helicity state does not exist without reference to some particular boost direction as
noted in footnote 6.
Using eqs. (2.72) and (2.73), with SµR = (0 ; s), the following two important formulae are
obtained:
√p·σ S ·σ√p·σ = m~σ ·s , (3.17)
√p·σ S ·σ
√p·σ = −m~σ ·s . (3.18)
These results can also be derived directly by employing the explicit form for the spin vector Sµ
[eq. (3.15)] and the results of eqs. (2.69) and (2.70).
The two-component spinor wave functions x and y can now be given explicitly in terms
of the χs defined in eq. (B.6). First, we note that eq. (3.9) when evaluated in the rest frame
yields x1 = y1 and x2 = y2. That is, as column vectors, xα(~p = 0) = yα(~p = 0) can be
expressed in general as some linear combination of the χs (s = ±12). Hence, we may choose
xα(~p = 0, s) = yα(~p = 0, s) =√mχs, where the factor of
√m reflects the standard relativistic
normalization of the rest-frame spin states. These wave functions can be boosted to an arbitrary
frame using eq. (2.68). The resulting undotted spinor wave functions are given by (see [62]) for
related expressions
xα(~p, s) =√p·σ χs , xα(~p, s) = −2sχ†
−s√p·σ , (3.19)
yα(~p, s) = 2s√p·σ χ−s , yα(~p, s) = χ†
s
√p·σ , (3.20)
6Strictly speaking, p is not defined in the rest frame. In practice, helicity states are defined in some movingframe with momentum ~p. The rest frame is achieved by boosting in the direction of −~p.
16
and the dotted spinor wave functions are given by
xα(~p, s) = −2s√p·σ χ−s , xα(~p, s) = χ†
s
√p·σ , (3.21)
yα(~p, s) =√p·σ χs , yα(~p, s) = 2sχ†
−s√p·σ , (3.22)
where√p·σ and
√p·σ are defined in eqs. (2.69) and (2.70).
The phase choices in eqs. (3.19)–(3.22) are consistent with those employed for four-component
spinor wave functions [see Appendix E]. We again emphasize that in eqs. (3.19)–(3.22), one may
either choose χs to be an eigenstate of ~σ ·s, where the spin is measured in the rest frame along
the quantization axis s, or choose χs to be an eigenstate of ~σ ·p (in this case we write s = λ),
which yields the helicity spinor wave functions.
The following equations can now be derived:
(S ·σ)αβxβ(~p, s) = 2syα(~p, s) , (S ·σ)αβ yβ(~p, s) = −2sxα(~p, s) , (3.23)
(S ·σ)αβ xβ(~p, s) = −2syα(~p, s) , (S ·σ)αβyβ(~p, s) = 2sxα(~p, s) , (3.24)
xα(~p, s)(S ·σ)αβ = −2syβ(~p, s) , yα(~p, s)(S ·σ)αβ = 2sxβ(~p, s) , (3.25)
xα(~p, s)(S ·σ)αβ = 2syβ(~p, s) , yα(~p, s)(S ·σ)αβ = −2sxβ(~p, s) . (3.26)
For example, using eqs. (3.17) and (3.18) and the definitions above for xα(~p, s) and yα(~p, s), we
find (suppressing spinor indices),
√p·σ S ·σ x(~p, s) =
√p·σ S ·σ√p·σ χs = m~σ ·sχs = 2smχs . (3.27)
Multiplying both sides of eq. (3.27) by√p·σ and noting that
√p·σ√p·σ = m, we end up with
S ·σ x(~p, s) = 2s√p·σ χs = 2sy(~p, s) . (3.28)
All the results of eqs. (3.23)–(3.26) can be derived in this manner.
The consistency of eqs. (3.23)–(3.26) can also be checked as follows. First, each of these
equations yields
(S ·σ)αα(S ·σ)αβ = −δβα , (S ·σ)αα(S ·σ)αβ = −δαβ. (3.29)
after noting that 4s2 = 1 (for s = ± 12). From eqs. (2.39) and (2.40) it follows that S ·S = −1,
as required. Second, if one applies
(p·σ S ·σ + S ·σ p·σ)αβ = 2p·S δαβ , (p·σ S ·σ + S ·σ p·σ)αβ = 2p·S δαβ , (3.30)
to eqs. (3.9)–(3.12) and eqs. (3.23)–(3.26), it follows that p·S = 0.
It is useful to combine the results of eqs. (3.9)–(3.12) and eqs. (3.23)–(3.26) as follows:
(pµ − 2smSµ)σαβµ xβ(~p, s) = 0 , (pµ − 2smSµ)σµ
αβxβ(~p, s) = 0 , (3.31)
(pµ + 2smSµ)σαβµ yβ(~p, s) = 0 , (pµ + 2smSµ)σµ
αβyβ(~p, s) = 0 , (3.32)
xα(~p, s)σµαβ
(pµ − 2smSµ) = 0 , xα(~p, s)σαβµ (pµ − 2smSµ) = 0 , (3.33)
yα(~p, s)σµαβ
(pµ + 2smSµ) = 0 , yα(~p, s)σαβµ (pµ + 2smSµ) = 0 . (3.34)
17
Eqs. (3.23)–(3.26) and eqs. (3.31)–(3.34) also apply to the helicity wave functions x(~p, λ) and
y(~p, λ) simply by replacing s with λ and Sµ(s) [eq. (3.15)] with Sµ(p) [eq. (3.16)].
The above results are applicable only for massive fermions (where the spin four-vector Sµ
exists). We may treat the case of massless fermions directly by employing helicity spinors in
eqs. (3.19)–(3.22). Putting E = |~p| and m = 0, we easily obtain:
xα(~p, λ) =√
2E (12 − λ)χλ , xα(~p, λ) =
√2E (1
2 − λ)χ†−λ , (3.35)
yα(~p, λ) =√
2E (12 + λ)χ−λ , yα(~p, λ) =
√2E (1
2 + λ)χ†λ , (3.36)
or equivalently
xα(~p, λ) =√
2E (12 − λ)χ−λ , xα(~p, λ) =
√2E (1
2 − λ)χ†λ , (3.37)
yα(~p, λ) =√
2E (12 + λ)χλ , yα(~p, λ) =
√2E (1
2 + λ)χ†−λ . (3.38)
It follows that:
(12 + λ
)x(~p, λ) = 0 ,
(12 + λ
)x(~p, λ) = 0 , (3.39)
(12 − λ
)y(~p, λ) = 0 ,
(12 − λ
)y(~p, λ) = 0 , (3.40)
The significance of eqs. (3.39) and (3.40) is clear; for massless fermions, only one helicity com-
ponent of x and y is non-zero. Applying this result to neutrinos, we find that massless neutrinos
are left-handed (λ = −1/2), while anti-neutrinos are right-handed (λ = +1/2).
Eqs. (3.39) and (3.40) can also be derived by carefully taking the m→ 0 limit of eqs. (3.31)
and (3.32) applied to the helicity wave functions x(~p, λ) and y(~p, λ) [i.e., replacing s with λ].
We then replace mSµ with pµ, which is the leading term in the limit of E m. Using the
results of eqs. (3.9) and (3.10) and dividing out by an overall factor or m (before finally taking
the m→ 0 limit) reproduces eqs. (3.39) and (3.40).
Having defined explicit forms for the two-component spinor wave functions, we can now
write down the spin projection matrices. Noting that 12 (1+2s~σ ·s)χs′ = 1
2(1+4ss′)χs′ = δss′χs′
(since s, s′ = ±12), one can write:
χsχ†s
= 12 (1 + 2s~σ ·s)
∑
s′
χs′χ†s′. (3.41)
Using the completeness relation given in eq. (B.8), and eq. (3.17) for ~σ ·s, it follows that
χsχ†s
= 12
(1 +
2s
m
√p·σ S ·σ√p·σ
), (3.42)
18
Hence, with both spinor indices in the lowered position,
x(~p, s)x(~p, s) =√p·σ χsχ†
s
√p·σ
= 12
√p·σ
[1 +
2s
m
√p·σS ·σ√p·σ
]√p·σ
= 12
[p·σ +
2s
mp·σS ·σp·σ
]
= 12 [p·σ − 2smS ·σ] . (3.43)
In the final step above, we simplified the product of three dot-products by noting that p·S = 0
implies that S ·σ p·σ = −p·σ S ·σ. The other spin projection formulae for massive fermions can
be similarly derived. The complete set of such formulae is given below: (see also [62])
xα(~p, s)xβ(~p, s) = 12(pµ − 2smSµ)σ
µ
αβ, (3.44)
yα(~p, s)yβ(~p, s) = 12(pµ + 2smSµ)σαβµ , (3.45)
xα(~p, s)yβ(~p, s) = 1
2
(mδα
β − 2s[S ·σ p·σ]αβ), (3.46)
yα(~p, s)xβ(~p, s) = 12
(mδαβ + 2s[S ·σ p·σ]αβ
), (3.47)
or equivalently,
xα(~p, s)xβ(~p, s) = 12 (pµ − 2smSµ)σαβµ , (3.48)
yα(~p, s)yβ(~p, s) = 12 (pµ + 2smSµ)σ
µ
αβ, (3.49)
yα(~p, s)xβ(~p, s) = − 1
2
(mδα
β + 2s[S ·σ p·σ]αβ), (3.50)
xα(~p, s)yβ(~p, s) = − 12
(mδαβ − 2s[S ·σ p·σ]αβ
). (3.51)
For the case of massless spin-1/2 fermions, we must use helicity spinor wave functions. The
corresponding massless projection operators can be obtained directly from the explicit forms for
the two-component spinor wave functions given in eqs. (3.35)–(3.38):
xα(~p, λ)xβ(~p, λ) = ( 12 − λ)p·σαβ , xα(~p, λ)xβ(~p, λ) = ( 1
2 − λ)p·σαβ , (3.52)
yα(~p, λ)yβ(~p, λ) = ( 12 + λ)p·σαβ , yα(~p, λ)yβ(~p, λ) = ( 1
2 + λ)p·σαβ , (3.53)
xα(~p, λ)yβ(~p, λ) = 0 , yα(~p, λ)xβ(~p, λ) = 0 , (3.54)
yα(~p, λ)xβ(~p, λ) = 0 , xα(~p, λ)yβ(~p, λ) = 0 . (3.55)
As a check, one can verify that the above results follow from eqs. (3.44)–(3.51), by replacing s
with λ, setting mSµ = pµ, and taking the m→ 0 limit at the end of the computation.
Having listed the projection operators for definite spin projection or helicity, we may now
sum over spins to derive the spin-sum identities. These arise when computing squared matrix
elements for unpolarized scattering and decay. There are only four basic identities, but for
19
convenience we list each of them with the two index height permutations that can occur in
squared amplitudes by following the rules given in this paper. The results can be derived by
inspection of the spin projection operators, since summing over s = ± 12 simply removes all terms
linear in the spin four-vector Sµ.
∑
s
xα(~p, s)xβ(~p, s) = p·σαβ ,∑
s
xα(~p, s)xβ(~p, s) = p·σαβ , (3.56)
∑
s
yα(~p, s)yβ(~p, s) = p·σαβ ,∑
s
yα(~p, s)yβ(~p, s) = p·σαβ , (3.57)
∑
s
xα(~p, s)yβ(~p, s) = mδαβ ,
∑
s
yα(~p, s)xβ(~p, s) = −mδαβ , (3.58)
∑
s
yα(~p, s)xβ(~p, s) = mδαβ ,∑
s
xα(~p, s)yβ(~p, s) = −mδαβ . (3.59)
These results are applicable both to spin-sums and helicity-sums, and hold for both massive and
massless spin-1/2 fermions.
One can also work out generalizations of the massive and massive projection operators.
These are products of two-component spinor wave functions, where the spin or helicity of each
spinor may be different. These are the Bouchiat-Michel formulae [135], which are derived in
Appendix F.
3.2 Fermion mass diagonalization and external wave functions in a generaltheory
Consider a collection of free anti-commuting two-component spin-1/2 fields, ξαi(x), which trans-
form as ( 12 , 0) fields under the Lorentz group. Here, α is the spinor index, and i labels the
distinct fields of the collection. The free-field Lagrangian is given by (see for example [67] for a
discussion of this Lagrangian)
L = i¯ξ iσµ∂µξi − 1
2Mij ξiξj − 1
2Mij¯ξ i
¯ξ j , (3.60)
where
Mij ≡ (M ij)∗. (3.61)
Note thatM ij is a complex symmetric matrix, since the product of anticommuting two-component
fields satisfies ξiξj = ξj ξi [with the spinor contraction rule according to eq. (2.25)].
In eq. (3.60), we have used the following convention concerning the “flavor” labels i and j.
Each left-handed ( 12 , 0) fermion always has an index with the opposite height of the corresponding
right-handed (0, 12 ) fermion. Raised indices can only be contracted with lowered indices and vice
versa. Flipping the heights of all flavor indices of an object corresponds to complex conjugation,
20
as in eq. (3.61).7
We can diagonalize the mass matrix and rewrite the Lagrangian in terms of mass eigenstates
ξαi , which have corresponding real non-negative masses mi. To do this, we introduce a unitary
matrix Ω
ξi = Ωikξk (3.62)
and demand that M ijΩikΩj
` = mkδk` (no sum over k), where the mk are real and non-negative.
Equivalently, in matrix notation with suppressed indices,8
ΩTM Ω = m = diag(m1,m2, . . .). (3.63)
This is the so-called Takagi diagonalization [69, 70] of an arbitrary complex symmetric matrix,
which is discussed in more detail in Appendix D. To compute the values of the diagonal elements
of m, note that
ΩM †MΩ† = m2 . (3.64)
Indeed M †M is hermitian and thus it can be diagonalized by a unitary matrix. Hence, the
elements of the diagonal matrix m are the non-negative square roots of the corresponding
eigenvalues of M †M . However, in cases where M †M has degenerate eigenvalues, eq. (3.64)
cannot be employed to determine the unitary matrix Ω that satisfies eq. (3.63). A more general
technique for determining Ω that works in all cases is given in Appendix D.
In terms of the mass eigenstates,
L = iξiσµ∂µξi − 12mi(ξiξi + ξiξi) . (3.65)
Each ξαi can now be expanded in a Fourier series, exactly as in the previous subsection:
ξαi(x) =∑
s
∫d3~p
(2π)3/2(2Eip)1/2
[xαi(~p, s)ai(~p, s)e
−ip·x + yαi(~p, s)a†i (~p, s)e
ip·x], (3.66)
where Eip ≡ (|~p|2 + m2i )
1/2, and the creation and annihilation operators, a†i and ai satisfy
anticommutation relations:
ai(~p, s), a†j(~p ′, s′) = δ3(~p− ~p ′)δss′δij . (3.67)
We employ covariant normalization of the one particle states, i.e., we act with one creation
operator on the vacuum with the following convention
|~p, s〉 ≡ (2π)3/2(2Eip)1/2a†i (~p, s) |0〉 , (3.68)
7In the case at hand, we have more specifically chosen all of the left-handed fermions to have lowered flavorindices, which implies that all of the right-handed fermions have raised flavor indices. However, in cases where asubset of left-handed fermions transform according to some representation R of a (global) symmetry whereas adifferent subset of left-handed fermions transform according to the conjugate representation R∗, it is often moreconvenient to employ a raised flavor index for the latter subset of left-handed fields.
8In general, the mi are not the eigenvalues of M . Rather, they are the singular values of the matrix M , whichare defined to be the positive square roots of the eigenvalues of M †M . See Appendix D for further details.
21
so that⟨~p|~p ′⟩ = (2π)3(2Eip)δ3(~p− ~p ′).
There is a useful modification to the mass diagonalization procedure above that is convenient
when there are massive Dirac fermions carrying a conserved charge. The key observation is that
one only needs a diagonal squared-mass matrix to ensure that the denominators of propagators
are diagonal. If χα is a charged massive field, then there must be an associated independent
two-component spinor field ηα of equal mass with the opposite charge. They appear in the
free-field Lagrangian as [6]:
L = iχσµ∂µχ+ iησµ∂µη −m(χη + χη) . (3.69)
Together, χ and η constitute a single Dirac fermion. We can then write:
χα(x) =∑
s
∫d3~p
(2π)3/2(2Ep)1/2
[xα(~p, s)a(~p, s)e
−ip·x + yα(~p, s)b†(~p, s)eip·x], (3.70)
ηα(x) =∑
s
∫d3~p
(2π)3/2(2Ep)1/2
[xα(~p, s)b(~p, s)e
−ip·x + yα(~p, s)a†(~p, s)eip·x
], (3.71)
where Ep ≡ (|~p|2 + m2)1/2, the creation and annihilation operators, a†, b†, a and b satisfy
anticommutation relations:
a(~p, s), a†(~p ′, s′) = b(~p, s), b†(~p ′, s′) = δ3(~p− ~p ′)δs,s′ , (3.72)
and all other anticommutators vanish. We now must distinguish between two types of one
particle states, which we can call fermion (F ) and anti-fermion (A):
|~p, s;F 〉 ≡ (2π)3/2(2Ep)1/2a†(~p, s) |0〉 , (3.73)
|~p, s;A〉 ≡ (2π)3/2(2Ep)1/2b†(~p, s) |0〉 . (3.74)
Note that both η(x) and χ(x) can create |~p, s;F 〉 from the vacuum, while η(x) and χ(x) can
create |~p, s;A〉. The one-particle wave functions are given by:
〈0|χα(x) |~p, s;F 〉 = xα(~p, s)e−ip·x , 〈0| ηα(x) |~p, s;F 〉 = yα(~p, s)e−ip·x , (3.75)
〈F ; ~p, s| ηα(x) |0〉 = yα(~p, s)eip·x , 〈F ; ~p, s| χα(x) |0〉 = xα(~p, s)eip·x , (3.76)
〈0| ηα(x) |~p, s;A〉 = xα(~p, s)e−ip·x , 〈0| χα(x) |~p, s;A〉 = yα(~p, s)e−ip·x , (3.77)
〈A; ~p, s|χα(x) |0〉 = yα(~p, s)eip·x , 〈A; ~p, s| ηα(x) |0〉 = xα(~p, s)eip·x , (3.78)
and the eight other single-particle matrix elements vanish.
More generally, consider a collection of such free anti-commuting charged massive spin-
1/2 fields, which can be represented by pairs of two-component fields χαi(x), ηiα(x). These
fields transform in (possibly reducible) representations of the unbroken symmetry group that
22
are complex conjugates of each other. (This is the reason for the difference in the flavor index
height i.) The free-field Lagrangian is given by
L = i ¯χ iσµ∂µχi + i ¯η iσµ∂µη
i −M ijχiη
j −Mij ¯χ i ¯η j , (3.79)
where M ij is an arbitrary complex matrix, and Mi
j ≡ (M ij)
∗ as before. We diagonalize the
mass matrix by introducing eigenstates χi and ηi and unitary matrices L and R,
χi = Likχk , ηi = Rikη
k , (3.80)
and demand that M ijLi
kRj` = mkδk` (no sum over k). In matrix form, this is written as (see
footnote 8):
LTMR = M = diag(M1,M2, . . .), (3.81)
with the mi real and non-negative. The singular-value decomposition of linear algebra, discussed
more fully in Appendix D, states that for any complex matrix M , unitary matrices L and R
exist such that eq. (3.81) is satisfied. It follows that:9
LT(MM †)L∗ = R†(M †M)R = M 2. (3.82)
That is, since MM † andM †M are both hermitian (with the same real non-negative eigenvalues),
they can be diagonalized by unitary matrices. The diagonal elements of M are therefore the
non-negative square roots of the corresponding eigenvalues of MM † (or M †M).
Thus, in terms of the mass eigenstates,
L = iχiσµ∂µχi + iηiσµ∂µη
i −mi(χiηi + χiηi) . (3.83)
The mass matrix now consists of 2 × 2 blocks(
0 mimi 0
)along the diagonal. More importantly,
the squared-mass matrix is diagonal with doubly degenerate entries m2i that will appear in the
denominators of the propagators of the theory. It describes a collection of Dirac fermions.10
Therefore, the result of the mass diagonalization procedure in a general theory always
consists of a collection of Majorana fermions as in equation (3.65), plus a collection of Dirac
fermions as in equation (3.83). This is the basis of the Feynman rules to be presented in the
next section.
For completeness, we review the squared-mass matrix diagonalization procedure for scalar
fields. First, consider a collection of free commuting real spin-0 fields, ϕi(x), where the flavor
9Consistency of notation requires that (M †)ij = Mji = (M j
i)∗ [and likewise (M†)i
j = M ji = (Mj
i)∗]. Thispermits the multiplication of MM † and M†M in a U(N)-covariant fashion.
10Of course, one could always choose instead to treat the Dirac fermions in a basis with a fully diagonalizedmass matrix, as in equation (3.65), by defining ξ2i−1 = (χi+ηi)/
√2 and ξ2i = i(χi−ηi)/
√2. These fermion fields
do not carry well-defined charges, and are analogous to writing a charged scalar field φ and its oppositely-chargedconjugate φ∗ in terms of their real and imaginary parts. However, it is rarely, if ever, convenient to do so; practicalcalculations only require that the squared-mass matrix is diagonal, and it is of course more pleasant to employfields that carry well-defined charges.
23
index i again labels the distinct scalar fields of the collection. The free-field Lagrangian is given
by
L = 12∂µϕi∂
µϕi − 12M
2ijϕiϕj , (3.84)
where M 2ij is a real symmetric matrix. We diagonalize the scalar squared-mass matrix by
introducing mass-eigenstates ϕi and the orthogonal matrix Q such that ϕi = Qijϕj, with
M2ijQikQj` = m2
kδk` (no sum over k). In matrix form, the latter reads
QTM2Q = m2 = diag(m21,m
22, . . .) . (3.85)
This is the standard diagonalization problem for a real symmetric matrix. The eigenvalues m2k
are real.11
Second, consider a collection of free commuting complex spin-0 fields, Φi(x). For complex
fields, we follow the convention for flavor indices enunciated below eq. (3.61) [e.g., Φi = (Φi)∗].
The free-field Lagrangian is given by
L = ∂µΦi∂µΦi − (M2)ijΦiΦ
j , (3.86)
where (M 2)ij is an hermitian matrix [which satisfies (M 2)ij = (M2)ji (see footnote 9)].
We diagonalize the scalar squared-mass matrix by introducing mass-egienstates Φi and the
unitary matrix W such that Φi = WikΦk (and Φi = W i
kΦk), with (M 2)ijWi
kW j` = M2
k δk` (no
sum over k). In matrix form, the latter reads
W †M2W = M2 = diag(M 21 ,M
22 , . . .) . (3.87)
This is the standard diagonalization problem for an hermitian matrix. The eigenvalues m2k are
real (see footnote 11).
4 Feynman rules with two-component spinors
In order to systematically perform perturbative calculations using two-component spinors, we
here present the basic Feynman rules. The Feynman rules for some specific models are given
in the Appendices E, F and G. Two-component Feynman rules have also been discussed in
[24, 64–66]
4.1 External fermion rules
Let us consider a general theory, for which we may assume that the mass matrix for fermions has
been diagonalized as discussed in the previous section. The rules for assigning two-component
external state spinors are then as follows.12
11Negative eigenvalues of M2 imply that the naive vacuum is unstable. One should shift the scalar fields bytheir vacuum expectation values and check that the resulting scalar squared-matrix possesses only non-negativeeigenvalues.
12We will often suppress the momentum and spin arguments of the spinor wave functions.
24
• For an initial-state left-handed ( 12 , 0) fermion: x.
• For an initial-state right-handed (0, 12 ) fermion: y.
• For a final-state left-handed ( 12 , 0) fermion: x.
• For a final-state right-handed (0, 12) fermion: y.
Note that, in general, the two-component external state fermion wave functions are distinguished
by their Lorentz group transformation properties, rather than by their particle or antiparticle
status as in four-component Feynman rules. This helps to explain why two-component notation
is especially convenient for (i) theories with Majorana particles, in which there is no fundamental
distinction between particles and antiparticles, and (ii) theories like the Standard Model and
MSSM in which the left and right-handed fermions transform under different representations of
the gauge group and (iii) problems with polarized particle beams. These rules are summarized
in the mnemonic diagram of Figure 1.
x x
y y
L (12 , 0) fermion
R (0, 12) fermion
Initial State Final State
Figure 1: The external wave-function spinors should be assigned as indicated here, for initial-state and final-state left-handed ( 1
2 , 0) and right-handed (0, 12) fermions.
In contrast to four-component Feynman rules, the direction of the arrows do not correspond
to the flow of charge or fermion number. These rules simply correspond to the formulae for the
one-particle wave functions given in eqs. (3.7) and (3.8) [with the convention that |~p, s〉 is an
initial-state fermion and 〈~p, s| is a final-state fermion]. In particular, the arrows indicate the
spinor index structure, with fields of undotted indices flowing into any vertex and fields of dotted
indices flowing out of any vertex.
The rules above apply to any mass eigenstate two-component fermion external wave func-
tions. It is noteworthy that the same rules apply for the two-component fermions governed by
the Lagrangians of eq. (3.65) [Majorana] and eq. (3.83) [Dirac].
25
4.2 Propagators
Next we turn to the subject of fermion propagators for two-component fermions. A derivation of
the two-component fermion propagators using path integral techniques is given in Appendix C.
Here, we will follow the more elementary approach typically given in an initial textbook treat-
ment of quantum field theory.
Fermion propagators are the Fourier transforms of the free-field vacuum expectation values
of time-ordered products of two fermion fields. They are obtained by inserting the free-field
expansion of the two-component fermion field and evaluating the spin sums using the formulas
given in eqs. (3.56) and (3.59). For the case of a single neutral two-component fermion field ξ
of mass m [see eqs. (3.65)-(3.68)] [24, 46, 64–66, 68],
〈0| Tξα(x)ξβ(y) |0〉FT =i
p2 −m2 + iε
∑
s
xα(~p, s)xβ(~p, s) =i
p2 −m2 + iεp·σαβ , (4.1)
〈0| T ξα(x)ξβ(y) |0〉FT =i
p2 −m2 + iε
∑
s
yα(~p, s)yβ(~p, s) =i
p2 −m2 + iεp·σαβ , (4.2)
〈0| T ξα(x)ξβ(y) |0〉FT =i
p2 −m2 + iε
∑
s
yα(~p, s)xβ(~p, s) =i
p2 −m2 + iεmδαβ , (4.3)
〈0| Tξα(x)ξβ(y) |0〉FT =i
p2 −m2 + iε
∑
s
xα(~p, s)yβ(~p, s) =
i
p2 −m2 + iεmδα
β , (4.4)
where FT indicates the Fourier transform from position to momentum space.13 These results
have an obvious diagrammatic representation, as shown in Fig. 2.
(a) (b)
p
αβ
p
β α
ip·σαβp2 −m2
ip·σαβp2 −m2
(c) (d)β α αβ
im
p2 −m2δαβ
im
p2 −m2δαβ
Figure 2: Feynman rules for propagator lines of a neutral two-component fermion with mass m.(The +iε terms in the denominators have been omitted here and from now on, for simplicity.)
13The Fourier transform of a translationally invariant function f(x, y) ≡ f(x − y) is given by
f(x, y) =
Zd4p
(2π)4bf (p) e−ip·(x−y) .
In the notation of the text above, f(x, y)FT ≡ bf(p).
26
β α
p ip·σαβp2 −m2
or−ip·σβαp2 −m2
Figure 3: This rule summarizes the results of both figs. 2(a) and (b) for a neutral two-component fermion with mass m.
Note that the direction of the momentum flow pµ here is determined by the creation operator
that appears in the evaluation of the free-field propagator. Arrows on fermion lines always run
away from dotted indices at a vertex and toward undotted indices at a vertex.
There are clearly two types of fermion propagators. The first type preserves the direction of
arrows, so it has one dotted and one undotted index. For this type of propagator, it is convenient
to establish a convention where pµ in the diagram is defined to be the momentum flowing in the
direction of the arrow on the fermion propagator. With this convention, the two rules above for
propagators of the first type can be summarized by one rule, as shown in Fig. 3. Here the choice
of the σ or the σ version of the rule is uniquely determined by the height of the indices on the
vertex to which the propagator is connected.14 These heights should always be chosen so that
they are contracted as in eq. (2.25). It should be noted that in diagrams (a) and (b) of Fig. 2
as drawn, the indices on the σ and σ read from right to left. This means that the most efficient
way to use the propagator rules of diagrams (a) and (b) [or equivalently, the propagator rule of
Fig. 3] in a Feynman diagram computation is to traverse the propagator lines in the direction
antiparallel [parallel] to the arrowed line segment for the σ [σ] version of the rule.
The second type of propagator shown in diagrams (c) and (d) of Fig. 2 does not preserve the
direction of arrows, and corresponds to an odd number of mass insertions. The indices on δ αβ
and δαβ are staggered as shown to indicate that α or α are to be contracted with an expression
to the left, while β or β are to be contracted with an expression to the right, in accord with
eq. (2.25).15
Starting with massless fermion propagators, one can derive the massive fermion propagators
by employing mass insertions as interaction vertices, as shown in Fig. 4. By summing up an
infinite chain of such mass insertions between massless fermion propagators, one can reproduce
the massive fermion propagators of both types.
It is convenient to treat separately the case of charged massive fermions. Consider a charged
Dirac fermion of massm, which is described by a pair of two-component fields χ and η [eq. (3.69)].
Using the free field expansions given by eqs. (3.70) and (3.71), and the appropriate spin-sums
14The second form of the rule in Fig. 3 arises when when one flips diagram (b) of Fig. 2 around by a 180
rotation (about an axis perpendicular to the plane of the diagram), and then relabels p → −p, α → β and β → α.15As in Fig. 3, alternative versions of the rules corresponding to diagrams (c) and (d) of Fig. 2 can be given
for which the indices on the Kronicker deltas are staggered as δβ α and δβα. These versions correspond to flipping
the two respective diagrams by 180 and relabeling the indices α → β and β → α.
27
β α× ×
αβ
−imδαβ −imδαβ
Figure 4: Fermion mass insertions (indicated by the crosses) can be treated as a type ofinteraction vertex, using the Feynman rules shown here.
[eqs. (3.56)–(3.59)], the two-component free-field propagators are obtained:
〈0| Tχα(x)χβ(y) |0〉FT = 〈0| Tηα(x)ηβ(y) |0〉FT =i
p2 −m2p·σαβ , (4.5)
〈0| T χα(x)χβ(y) |0〉FT = 〈0|T ηα(x)ηβ(y) |0〉FT =i
p2 −m2p·σαβ , (4.6)
〈0| Tχα(x)ηβ(y) |0〉FT = 〈0| Tηα(x)χβ(y) |0〉FT =i
p2 −m2mδα
β , (4.7)
〈0| T χα(x)ηβ(y) |0〉FT = 〈0| T ηα(x)χβ(y) |0〉FT =i
p2 −m2mδαβ . (4.8)
For all other combinations of fermion bilinears, the corresponding two-point functions vanish.
These results again have a simple diagrammatic representation, as shown in Fig. 5.
(a) (b)χ χ ηη
p
αβ β α
p
ip·σαβp2 −m2
or−ip·σβαp2 −m2
ip·σαβp2 −m2
or−ip·σβαp2 −m2
χ η ηχ(c) (d)β α αβ
im
p2 −m2δαβ
im
p2 −m2δαβ
Figure 5: Feynman rules for propagator lines of a pair of charged two-component fermions witha Dirac mass m. As in Fig. 3, the direction of the momentum is taken to flow from the dottedto the undotted index in diagrams (a) and (b).
Note that for Dirac fermions, the propagators with opposing arrows (proportional to a mass)
necessarily change the identity (χ or η) of the two-component fermion, while the single-arrow
propagators are diagonal in the fields. In processes involving such a charged fermion, one must
of course distinguish between the χ and η fields.
For completeness, we provide the propagators for scalar and vector bosons in Fig. 6.
28
i
p2 −m2
µ ν
−ip2 −m2
[gµν + (ξ − 1)
pµpν
p2 − ξm2
]
Figure 6: The Feynman rules for propagators of scalar bosons, and vector bosons in Rξ gauge,carrying momentum pµ in each case. Here ξ = 1 is Feynman gauge, and ξ = 0 is Landau gauge.
4.3 Fermion interactions with bosons
We next discuss the interaction vertices for fermions with bosons. Renormalizable Lorentz-
invariant interactions involving fermions must consist of bilinears in the fermion fields, which
transform as a Lorentz scalar or vector, coupled to the appropriate bosonic scalar or vector field
to make an overall Lorentz scalar quantity.
Let us write all of the two-component left-handed fermions of the theory as ψj, where j
runs over all of the gauge group representation and flavor degrees of freedom. The most general
set of interactions with the scalars of the theory φI are then given by:
Lint = −12 Y
IjkφI ψjψk − 12 YIjkφ
I ¯ψ j ¯ψ k , (4.9)
where YIjk = (Y Ijk)∗ and φI = (φI)∗. We have suppressed the spinor indices here; the product
of two component spinors is always performed according to the index convention indicated in
eq. (2.25). The flavor index I runs over a collection of real scalar fields ϕi and pairs of complex
scalar fields Φj and (Φj)∗.16 The Yukawa couplings Y Ijk are symmetric under interchange of j
and k. The hatted fields are the so-called interaction-eigenstate fields.
However, in general the mass-eigenstates can be different, as discussed in subsection 3.2.
The computation of matrix elements for physical processes is more conveniently done in terms
of the propagating mass-eigenstate fields. In general, the interaction-eigenstate ( 12 , 0)-fermion
fields ψi consist of Majorana fermions ξi, and Dirac fermion pairs χi and ηi after mass terms
(both explicit and coming from spontaneous symmetry breaking) are taken into account. The
mass-eigenstate basis ψ is related to the interaction-eigenstate basis ψ by a unitary rotation Uij
on the flavor indices. In matrix form:
ψ ≡
ξχη
= Uψ ≡
Ω 0 00 L 00 0 R
ξχη
, (4.10)
where Ω, L, and R are constructed as described previously in Section 3.2 [see eqs. (3.63) and
(3.81)]. Likewise, the interaction-eigenstate scalar fields φI generally consist of real scalar fields
16For example, in a theory with one complex scalar field Φ, we would take φ1 = Φ and φ2 = Φ∗.
29
I
k, β
j, α
−iY Ijkδαβ or − iY Ijkδβ
α(a)
I
k, β
j, α
−iYIjkδαβ or − iYIjkδβ α(b)
Figure 7: Feynman rules for Yukawa couplings of scalars to two-component fermions in ageneral field theory. The choice of which rule to use depends on how the vertex connects tothe rest of the amplitude. When indices are suppressed, the spinor index part is always justproportional to the identity matrix.
ϕi and complex scalar fields Φi. The mass eigenstate basis φ is related to the interaction
eigenstate basis φ by a unitary rotation VIJ on the flavor indices. In matrix form:
φ ≡(ϕ
Φ
)= V φ ≡
(Q 00 W
)(ϕΦ
), (4.11)
where Q and W are constructed according to eqs. (3.85) and (3.87).
Thus, we may rewrite eq. (4.9) in terms of mass-eigenstate fields:
Lint = −12Y
IjkφIψjψk − 12YIjkφ
I ψ jψ k , (4.12)
where
Y Ijk = VLIUm
jUnkY Lmn . (4.13)
The corresponding Feynman rules are shown in Fig. 7. Note that if the scalar φI is complex,
then one can associate an arrow with the flow of analyticity,17 which would point into the vertex
in (a) and would point out of the vertex in (b).
The renormalizable interactions of vector bosons with fermions and scalars arise from gauge
interactions. These interaction terms of the Lagrangian derive from the respective kinetic energy
terms of the fermions and scalars when the derivative is promoted to the covariant derivative:
(Dµ)ij ≡ δij∂µ + igaA
aµ(T
a)ij , (4.14)
where the index a labels the (real or complex) vector bosons Aµa and is summed over. The index
17As in the case of the fermions, the arrow on the dashed line representing the scalar field does not representthe flow of a conserved charge. It simply keeps track of the height of the scalar flavor index entering or leaving agiven vertex.
30
a runs over the adjoint representation of the gauge group,18 and the (T a)ij are hermitian rep-
resentation matrices19 of the Lie algebra of the gauge group acting on the left-handed fermions.
There is a separate coupling ga for each simple group or U(1) factor of the gauge group G.20
In the gauge-interaction basis for the left-handed two-component fermions the corresponding
interaction Lagrangian is given by
Lint = −gaAµa¯ψ i σµ(T
a)ijψj . (4.15)
In the case of spontaneously broken gauge theories, one must diagonalize the vector boson
squared mass matrix. The form of eq. (4.15) still applies where Aaµ are gauge boson fields of
definite mass, although in this case for a fixed value of a, gaTa [which multiplies Aaµ in eq. (4.15)]
is some linear combination of the original gaTa of the unbroken theory.21 Henceforth, we assume
that that the Aaµ are the gauge boson mass-eigenstate fields.
To obtain the desired Feynman rule, we must rewrite eq. (4.15) in terms of mass-eigenstate
fermion fields. The resulting interaction Lagrangian takes the form
Lint = −Aµa ψi σµ(Ga)ijψj , (4.17)
where
(Ga)ij = gaU
ki(T
a)kmUm
j , (4.18)
or in matrix form, Ga = gaU†T aU (no sum over a). Note that Ga is an hermitian matrix. The
corresponding Feynman rule is shown in Fig. 8.
The above treatment of gauge interactions of (two-component) fermions is general, but it
is useful to consider separately the special case of gauge interactions of charged Dirac fermions.
Consider pairs of left-handed ( 12 , 0) interaction-eigenstate fermions χi and ηi that transform as
conjugate representations of the gauge group (hence the difference in the flavor index heights).
The fermion mass matrix couples χ and η type fields as in eq. (3.79). The Lagrangian for the
gauge interactions of Dirac fermions can be written in the form:
Lint = −gaAµa ¯χ i σµ(Ta)i
jχj + gaAµa¯η i σµ(T
a)jiηj , (4.19)
18Since the adjoint representation is a real representation, the height of the adjoint index a is not significant.The choice of a subscript or superscript adjoint index is based solely on typographical considerations.
19For a U(1) gauge group, the T a are replaced by real numbers corresponding to the U(1) charges of theleft-handed ( 1
2, 0) fermions.
20That is, the generators T a separate out into distinct classes, each of which is associated with a simple groupor one of the U(1) factors contained in the direct product that defines G. In particular, ga = gb if T a and T b arein the same class. If G is simple, then ga = g for all a.
21For example, in the electroweak Standard Model, G=SU(2)×U(1) and T a = ( 12τa , 1
2Y ), where the τa are
the usual Pauli matrices. Then, after diagonalizing the gauge boson squared-mass matrix, one finds:
12gW a
µτa + 12g′BµY =
g
2√
2(W+
µ τ+ + W−µ τ−) +
g
2 cos θW
`τ 3 + 2Q sin2 θW
´Zµ + eQAµ , (4.16)
where τ± ≡ τ 1 ± iτ 2, Q = 12(τ 3 + Y ), and e = g sin θW = g′ cos θW . Here W a
µ , Bµ are the gauge fields of theunbroken theory and W±, Z and A are the gauge boson mass-eigenstates of the broken theory.
31
a, µj, β
i, α
−i(Ga)ij σαβµ or i(Ga)ij σµβα
Figure 8: The Feynman rules for two-component fermion interactions with vector bosons. Thechoice of which rule to use depends on how the vertex connects to the rest of the amplitude. Ga
is defined in eq. (4.18).
where the Aaµ are gauge boson mass-eigenstate fields. Here we have used the fact that if (T a)ij
are the representation matrices for the χi, then the ηi transform in the complex conjugate
representation with generator matrices −(T a)∗ = −(T a)T, where we have used the hermiticity
of the generator matrices. Again we rewrite eq. (4.19) in terms of mass-eigenstate fermion fields.
The resulting interaction Lagrangian is given by:
Lint = −Aµa χ i σµ(GaL)ijχj +Aµa η i σµ(G
aR)j
iηj , (4.20)
where
(GaL)ij = gaL
ki(T
a)kmLm
j , (4.21)
(GaR)ji = gaR
mj(T
a)mkRk
i . (4.22)
In matrix form, eqs. (4.21) and (4.22) read: GaL = gaL
†T aL and GaR = gaR†T aR (no sum
over a); GaL and GaR are hermitian matrices. The corresponding Feynman rules for the gauge
interactions of Dirac fermions are shown in Fig. 9.
a, µβ
α
−i(GaL)ij σαβµ or ig(GaL)i
j σµβα
χi
χj
a, µβ
α
i(GaR)ji σαβµ or −ig(GaR)j
i σµβα
ηi
ηj
Figure 9: The Feynman rules for two-component fermion interactions with vector bosons, inthe case that χi and ηi form a Dirac fermion. The matrices Ga
L and GaR are related to the groupgenerators for the representation carried by the χi according to eqs. (4.21) and (4.22). Thetwo-component field labels conform to the conventions of Section 5.
In Figs. 7–9, two versions are given for each of the boson-fermion-fermion Feynman rules.
The correct version to use depends in a unique way on the heights of indices used to connect
32
each fermion line to the rest of the diagram. For example, the way of writing the vector-fermion-
fermion interaction rule depends on whether we used ψiσµψj , or its equivalent form −ψjσµψi,from eq. (4.15). Note the different heights of the undotted and dotted spinor indices that adorn
σµ and σµ. The choice of which rule to use is thus dictated by the height of the indices on
the lines that connect to the vertex. These heights should always be chosen so that they are
contracted as in eq. (2.25). Similarly, for the scalar-fermion-fermion vertices, one should choose
the rule which correctly matches the indices with the rest of the diagram. However, when all
spinor indices are suppressed, the scalar-fermion-fermion rules will have an identical appearance
for both cases, since they are just proportional to the identity matrix on the 2×2 spinor space.)
These above comments will be clarified by examples in Section 4.5. Numerous examples
and applications of the results of this subsection can be found in Appendices D and E.
4.4 General structure and rules for Feynman graphs
When computing an amplitude for a given process, all possible diagrams should be drawn that
conform with the rules given above for external wavefunctions, propagators, and interactions.
Starting from any external wave function spinor, or from any vertex on a fermion loop, factors
corresponding to each propagator and vertex should be written down from left to right, following
the line until it ends at another external state wave function or at the original point on the
fermion loop. If one starts a fermion line at an x or y external state spinor, it should have
a raised undotted index in accord with eq. (2.25). Or, if one starts with an x or y, it should
have a lowered dotted spinor index. Then, all spinor indices should always be contracted as in
eq. (2.25). If one ends with an x or y external state spinor, it will have a lowered undotted index,
while if one ends with an x or y spinor, it will have a raised dotted index. For arrow-preserving
fermion propagators and gauge vertices, the preceding determines whether the σ or σ rule should
be used. With only a little practice, one can write down amplitudes immediately with all spinor
indices suppressed.
Symmetry factors for identical particles are implemented in the usual way. Fermi-Dirac
statistics are implemented by the following rules:
• Each closed fermion loop gets a factor of −1.
• A relative minus sign is imposed between terms contributing to a given amplitude whenever
the ordering of external state spinors (written left-to-right) differs by an odd permutation.
Amplitudes generated according to these rules will contain objects of the form:
a = z1Σz2 (4.23)
where z1 and z2 are each commuting external spinor wave functions x, x, y, or y, and Σ is a
33
sequence of alternating σ and σ matrices. The complex conjugate of this quantity is given by
a∗ = z2Σrz1 (4.24)
where Σr is obtained from Σ by reversing the order of all the σ and σ matrices, and using the
same rule for suppressed spinor indices. (Notice that this rule for taking complex conjugates
has the same form as for anticommuting spinors.) We emphasize that in principle, it does not
matter in what direction a diagram is traversed while applying the rules. However, for each
diagram one must include a sign that depends on the ordering of the external fermions. This
sign can be fixed by first choosing some canonical ordering of the external fermions. Then for any
graph that contributes to the process of interest, the corresponding sign is positive (negative)
if the ordering of external fermions is an even (odd) permutation with respect to the canonical
ordering. If one chooses a different canonical ordering, then the resulting amplitude changes
by an overall sign (is unchanged) if this ordering is an odd (even) permutation of the original
canonical ordering.22 This is consistent with the fact that the amplitude is only defined up to
an overall sign, which is not physically observable.
Note that different graphs contributing to the same process will often have different external
state wave function spinors, with different arrow directions, for the same external fermion.
Furthermore, there are no arbitrary choices to be made for arrow directions, as there are in
some four-component Feynman rules for Majorana fermions. Instead, one must add together all
Feynman graphs that obey the rules.
4.5 Basic examples of writing down diagrams and amplitudes
A few simple examples will help clarify these rules. (A larger number of examples, drawn from
practical calculations, are given in section 6.) Let us first consider a theory with a single,
uncharged, massive ( 12 , 0) fermion ξ, and a real scalar φ, with interaction
Lint = −12
(λξξ + λ∗ξξ
)φ. (4.25)
Consider the decay φ → ξ(~p1, s1)ξ(~p2, s2), where by ξ we mean the one particle state given by
eq. (3.6). Two diagrams contribute to this process, as shown in Figure 10.
The matrix element is then given by
iM = y(~p1, s1)α(−iλδαβ)y(~p2, s2)β + x(~p1, s1)α(−iλ∗δαβ)x(~p2, s2)
β
= −iλy(~p1, s1)y(~p2, s2)− iλ∗x(~p1, s1)x(~p2, s2). (4.26)
22For a process with exactly two external fermions, it is convenient to apply the Feynman rules by starting fromthe same fermion external state in all diagrams. That way, all terms in the amplitude have the same canonicalordering of fermions and there are no additional minus signs between diagrams. However, if there are four or moreexternal fermions, it may happen that there is no way to choose the same ordering of external state spinors forall graphs when the amplitude is written down. Then the relative signs between different graphs must be chosenaccording to the relative sign of the permutation of the corresponding external fermion spinors. This guaranteesthat the total amplitude is antisymmetric under the interchange of any pair of external fermions.
34
φ
ξ(p1, s1)
ξ(p2, s2)
φ
ξ(p1, s1)
ξ(p2, s2)
Figure 10: The two tree-level Feynman diagrams contributing to the decay of a scalar into aMajorana fermion pair.
The second line could be written down directly by recalling that the sum over suppressed spinor
indices is taken according to eq. (2.25). Note that if we reverse the ordering for the external
fermions, the overall sign of the amplitude changes sign. This is easily checked, since for the
commuting spinor wave functions (x and y), the spinor products in eq. (4.26) change sign when
the order is reversed [see eqs. (2.47) and (2.48)]. This overall sign is not significant and depends
on the order used in constructing the two particle state. One could even make the choice of
starting the first diagram from fermion 1, and the second diagram from fermion 2:
iM = −iλy(~p1, s1)y(~p2, s2)− (−1)iλ∗x(~p2, s2)x(~p1, s1) . (4.27)
Here the first term establishes the canonical ordering of fermions (12), and the contribution from
the second diagram therefore includes the relative minus sign in parentheses. Indeed, eqs. (4.26)
and (4.27) are equal. The computation of the total decay rate is straightforward. Of course,
one must multiply the integral over the total phase space by 1/2 to account for the identical
particles.
Consider next the decay of a massive neutral vector Aµ into a Majorana fermion pair
Aµ → ξ(~p1, s1)ξ(~p2, s2), following from the interaction
Lint = −GAµξσµξ , (4.28)
where G is a real coupling parameter. The two diagrams shown in Figure 11 contribute.
Aµ
ξ(p2, s2)
ξ(p1, s1)
Aµ
ξ(p2, s2)
ξ(p1, s1)
Figure 11: The two tree-level Feynman diagrams contributing to the decay of a massive vectorboson Aµ into a pair of Majorana fermions ξ.
We start from the fermion with momentum p1 and spin vector s1 and end at the fermion
with momentum p2 and spin vector s2, using the rules of Fig. 8. The resulting amplitude for
35
the decay is
iM = εµ [−iGx(~p1, s1)σµy(~p2, s2) + iGy(~p1, s1)σµx(~p2, s2)] (4.29)
where εµ is the vector boson polarization vector. We have used the σ-version of the vector-
fermion-fermion rule [see Fig. 8] for the first diagram of Fig. 11 and the σ-version for the
second diagram of Fig. 11, as dictated by the implicit spinor indices, which we have suppressed.
However, we could have chosen to evaluate the second diagram of Fig. 11 using the σ-version of
the vector-fermion-fermion rule by starting from the fermion with momentum p2. In that case,
the factor +iGy(~p1, s1)σµx(~p2, s2) in eq. (4.29) is replaced by
(−1)[−iGx(~p2, s2)σµy(~p1, s1)] . (4.30)
In eq. (4.30), the factor of −iG arises from the use of the σ-version of the vector-fermion-fermion
rule, and the overall factor of −1 appears because the order of the fermion wave functions has
been reversed; i.e. (21) is an odd permutation of (12). This is in accord with the ordering rule
stated at the end of Section 4.4. Thus, the resulting amplitude for the decay of the vector boson
into the pair of Majorana fermions now takes the form:
iM = εµ [−iGx(~p1, s1)σµy(~p2, s2) + iGx(~p2, s2)σµy(~p1, s1)] . (4.31)
By using yσµx = xσµy, which follows from eq. (2.49) with commuting spinors, one sees that
eqs. (4.29) and (4.31) are identical. The form given in eq. (4.31) explicitly exhibits the fact that
the amplitude is antisymmetric under the interchange of the two external identical fermions.
Again, the absolute sign of the total amplitude is not significant and depends on the choice of
ordering of the outgoing states.
Next, we consider the decay of a massive neutral vector boson into a charged fermion-
antifermion pair. Suppose that we identify χ and η as left-handed fields with charges Q = 1 and
Q = −1, respectively. The corresponding interaction is given by:
Lint = −Aµ[GL χ σµχ−GR η σµη ]. (4.32)
There are two contributing graphs, as shown in Figure 12.
Aµ
χ(p2, s2)
χ(p1, s1)
Aµ
η(p2, s2)
η(p1, s1)
Figure 12: The two tree-level Feynman diagrams contributing to the decay of a massive neutralvector boson Aµ into a Dirac fermion-antifermion pair.
To evaluate the amplitude, we start from the charge Q = +1 fermion (with momentum
p1 and spin vector s1), and end at the charge Q = −1 fermion (with momentum p2 and spin
36
vector s2). In particular, for the final state fermion lines, the outgoing χ with arrow pointing
outward from the vertex and the outgoing η with arrow pointing inward to the vertex both
correspond to outgoing Q = +1 states. The amplitude for the decay is
iM = εµ [−iGLx(~p1, s1)σµy(~p2, s2)− iGRy(~p1, s1)σµx(~p2, s2)]
= εµ [−iGLx(~p1, s1)σµy(~p2, s2)− iGRx(~p2, s2)σµy(~p1, s1)] . (4.33)
As in the case of the decay to a pair of Majorana fermions, we have exhibited two forms for the
amplitude in eq. (4.33) that depend on whether the σ-version or the σ-version of the Feynman
rule has been employed. Of course, the resulting amplitude is the same in each method (up to
an overall sign of the total amplitude which is not determined).
The next level of complexity consists of diagrams that involve fermion propagators. For our
first example of this type, consider the tree-level matrix element for the scattering of a neutral
scalar and a two-component neutral massive fermion (φξ → φξ), with the interaction Lagrangian
given above in eq. (4.25). Using the corresponding Feynman rules, there are eight contributing
diagrams. Four are depicted in Fig. 13; there are another four diagrams (not shown) where the
initial and final state scalars are crossed (i.e., the initial state scalar is attached to the same
vertex as the final state fermion).
k k
Figure 13: Tree-level Feynman diagrams contributing to the elastic scattering of a neutralscalar and a neutral two-component fermion. There are four more diagrams, obtained fromthese by crossing the initial and final scalar lines.
We shall write down the amplitudes for the four diagrams shown, starting with the final
state fermion line and moving toward the initial state fermion line. Then,
iM =−i
k2 −m2ξ
|λ|2 [x(~p2, s2)σ ·k x(~p1, s1) + y(~p2, s2)σ ·k y(~p1, s1)]
+mξ
[λ2y(~p2, s2)x(~p1, s1) + (λ∗)2x(~p2, s2)y(~p1, s1)
]+ (crossed) . (4.34)
37
k k
Figure 14: Tree-level Feynman diagrams contributing to the elastic scattering of a neutralvector boson and a neutral two-component fermion. There are four more diagrams, obtainedfrom these by crossing the initial and final scalar lines.
where kµ is the sum of the two incoming (or outgoing) four-momenta, (p1, s1) are the momentum
and spin four-vectors of the incoming fermion, and (p2, s2) are those of the outgoing fermion.
(We will not write down the “crossed” terms, which have the the initial and final scalars inter-
changed.) Note that we could have evaluated the diagrams above by starting with the initial
vertex and moving toward the final vertex. It is easy to check that the resulting amplitude is
the negative of the one obtained in eq. (4.34); the overall sign change simply corresponds to
swapping the order of the two fermions and has no physical consequence. The overall minus
sign is a consequence of eqs. (2.47)–(2.49) and the minus sign difference between the two ways
of evaluating the propagator that preserves the arrow direction.
Next, we compute the tree-level matrix element for the scattering of a vector boson and a
neutral massive two-component fermion ξ with the interaction Lagrangian of eq. (4.28). Again
there are eight diagrams: the four diagrams depicted in Fig. 14 plus another four (not shown)
where the initial and final state vector bosons are crossed. Starting with the final state fermion
line and moving toward the initial state, we obtain
iM =−iG2
k2 −m2ξ
x(~p2, s2)σ ·ε∗2 σ ·k σ ·ε1 x(~p1, s1) + y(~p2, s2)σ ·ε∗2 σ ·k σ ·ε1y (~p1, s1)
−mξ
[y(~p2, s2)σ ·ε∗2 σ ·ε1 x(~p1, s1) + x(~p2, s2)σ ·ε∗2 σ ·ε1 y(~p1, s1)
]+ (crossed) , (4.35)
where ε1 and ε2 are the initial and final vector boson polarization four-vectors, respectively.
As before, kµ is the sum of the two incoming (or outgoing) four-momenta, and (p1, s1) are the
momentum and spin four-vectors of the incoming fermion, and (p2, s2) are those of the outgoing
fermion. (We again omit the “crossed” terms, which have the the initial and final vector bosons
interchanged.) If one evaluates the diagrams above by starting with the initial vertex and moving
toward the final vertex, the resulting amplitude is the negative of the one obtained in eq. (4.35),
38
k
χ
η
χ
k
η
χ
η
χ
η χ
η η
χ η
χ
Figure 15: Tree-level Feynman diagrams contributing to the elastic scattering of a neutralscalar and a charged fermion. There are four more diagrams, obtained from these by crossingthe initial and final scalar lines.
as expected.
As our next example, we consider the scattering of a charged Dirac fermion with a neutral
scalar. The left-handed fields χ and η have opposite charges Q = +1 and −1 respectively, and
interact with the scalar φ according to
Lint = −φ[κχη + κ∗χη] , (4.36)
where κ is a coupling parameter. Then, for the elastic scattering of a Q = +1 fermion and a
scalar, the diagrams of Fig. 15 contribute at tree-level plus another four diagrams (not shown)
where the initial and final state scalars are crossed. Now, these diagrams match precisely those
of Fig. 13. Thus, applying the Feynman rules yields the same matrix element, eq. (4.34),
previously obtained for the scattering of a neutral scalar and neutral two-component fermion,
with the replacement of λ with κ.
Consider next the scattering of a charged Dirac fermion and a charged scalar, where both
the scalar and fermion have the same absolute value of the charge. As above, we denote the
charged Q = ±1 fermion by the pair of two-component fermions χ and η and the (intermediate
state) neutral two-component fermion by ξ. The charged Q = ±1 scalar is represented by the
scalar field φ and its complex conjugate. The interaction Lagrangian takes the form:
Lint = −φ∗[κ1χξ + κ∗2ηξ]− φ[κ∗1χξ + κ2ηξ] . (4.37)
Consider the scattering of an initial boson-fermion state into its charge-conjugated final state
via the exchange of a neutral fermion. The relevant diagrams are shown in Fig. 16 plus the
corresponding diagrams with the initial and final scalars crossed. We define the four-momentum
k to be the sum of the two initial state four-momenta as shown in Fig. 16. The derivation of
39
k
χ
ξ
η
k
η
ξ
χ
χ
ξ
χ η
ξ
η
Figure 16: Tree-level Feynman diagrams contributing to the scattering of an initial chargedscalar and a charged fermion into its charge-conjugated final state. The unlabeled intermediatestate is a neutral fermion. There are four more diagrams, obtained from these by crossing theinitial and final scalar lines.
the amplitude is similar to the ones given previously, and we end up with
iM =−i
k2 −m2ξ
κ1κ
∗2[x(~p2, s2)σ ·k x(~p1, s1) + y(~p2, s2)σ ·k y(~p1, s1)]
+mξ
[κ2
1y(~p2, s2)x(~p1, s1) + (κ∗2)2x(~p2, s2)y(~p1, s1)
]+ (crossed) . (4.38)
The scattering of a charged fermion and a neutral spin-1 vector boson can be similarly
treated. For example, consider the amplitude for the elastic scattering of a charged fermion
and a neutral vector boson. Again taking the interactions as given in eq. (4.32), the relevant
diagrams are those shown in Fig. 17, plus four diagrams (not shown) obtained from these by
crossing the initial and final state vectors.
k
χ
χ
χ
k
η
η
η
χ
χ η
η η
η χ
χ
Figure 17: Tree-level Feynman diagrams contributing to the elastic scattering of a neutralvector boson and a charged Dirac fermion. There are four more diagrams, obtained from theseby crossing the initial and final vector lines.
40
Applying the Feynman rules following from eq. (4.28) as before, one obtains the following
matrix element
iM =−i
k2 −m2ξ
G2Lx(~p2, s2)σ ·ε∗2 σ ·k σ ·ε1 x(~p1, s1) +G2
Ry(~p2, s2)σ ·ε∗2 σ ·k σ ·ε1y (~p1, s1)
−mGLGR[y(~p2, s2)σ ·ε∗2 σ ·ε1 x(~p1, s1) + x(~p2, s2)σ ·ε∗2 σ ·ε1 y(~p1, s1)
]+ (crossed) (4.39)
and the assignments of momenta and spins are as before.
The computation of the amplitude for the scattering of a charged fermion and a charged
vector boson is straightforward and will not be given explicitly here.
Finally, let us work out an example with four external-state fermions. Consider the case of
elastic scattering of two identical Majorana fermions due to scalar exchange, governed by the
interaction of eq. (4.25). The diagrams for scattering initial fermions labeled 1, 2 into final state
fermions labeled 3, 4 are shown in Fig. 18. The resulting matrix element is:
iM =−i
s−m2φ
λ2(x1x2)(y3y4) + (λ∗)2(y1y2)(x3x4) + |λ|2 [(x1x2)(x3x4) + (y1y2)(y3y4)]
+(−1)−i
t−m2φ
λ2(y3x1)(y4x2) + (λ∗)2(x3y1)(x4y2) + |λ|2 [(x3y1)(y4x2) + (y3x1)(x4y2)]
+−i
u−m2φ
λ2(y4x1)(y3x2) + (λ∗)2(x4y1)(x3y2) + |λ|2 [(x4y1)(y3x2) + (y4x1)(x3y2)]
,
(4.40)
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
Figure 18: Tree-level Feynman diagrams contributing to the elastic scattering of identicalneutral Majorana fermions via scalar exchange in the s-channel (top row), t-channel (middlerow), and u-channel (bottom row).
41
where xi ≡ x(~pi, si), yi ≡ y(~pi, si), mφ is the mass of the exchanged scalar, s = (p1 + p2)2,
t = (p1 − p3)2 and u = (p1 − p4)
2. The relative minus sign (in parentheses) between the t-
channel diagram and the s and u-channel diagrams is obtained by observing that 3142 is an odd
permutation and 4132 is an even permutation of 1234.23
Eq. (4.40) can be factorized with respect to the scalar line:
iM =−i
s−m2φ
(λx1x2 + λ∗y1y2)(λy3y4 + λ∗x3x4) +i
t−m2φ
(λy3x1 + λ∗x3y1)(λy4x2 + λ∗x4y2)
+−i
u−m2φ
(λy4x1 + λ∗x4y1)(λy3x2 + λ∗x3y2). (4.41)
This is a common feature of Feynman graphs with a virtual boson. This example also illustrates
the typical feature that, compared to the four-component formalism, two-component fermion
Feynman rules yield more diagrams, although the contribution of each of the diagrams is simpler.
4.6 Self-energy functions and pole masses for two-component fermions
In this section, we discuss the self-energy functions for fermions in two-component notation,
taking into account the possibilities of loop-induced mixing and absorptive parts corresponding
to decays to intermediate states. This formalism is useful in the computation of loop-corrected
physical pole masses.
Consider a theory with left-handed fermion degrees of freedom ψi labeled by an index
i = 1, 2, . . . , N . Associated which each ψi is a right-handed fermion¯ψi, where the flavor labels
are treated as described below eq. (3.61). The theory is assumed to contain arbitrary interactions,
which we will not need to refer to explicitly. As discussed in Section 3.2, we diagonalize the
fermion mass matrix and identify the fermion mass-eigenstates ψi as indicated in eq. (4.10).
In general, the mass-eigenstates consist of neutral Majorana fermions ξk (k = 1, . . . N − 2n)
and Dirac fermion pairs χ` and η` (` = 1, . . . , n).24 With respect to this basis, the symmetric
N×N tree-level fermion mass matrix, mij, is made up of diagonal elements mk and 2×2 blocks( 0 m`m` 0
)along the diagonal, where the mk and m` are real and non-negative. Since mij is real,
the height of the flavor indices is not significant. Nevertheless, it is useful to define mij ≡mij in
order to maintain the convention that two repeated flavor indices are summed when one index
is raised and the other is lowered.25 Note that mikmkj = mikmkj = m2
i δji is a diagonal matrix.
The full, loop-corrected Feynman propagators with four-momentum pµ are defined by the
Fourier transforms of vacuum expectation values of time-ordered products of bilinears of the
23Note that we would have obtained the same sign for the u-channel diagram had we crossed the initial statefermion lines instead of the final state fermion lines.
24In order to have a unified description, we shall take the flavor index of all left-handed fields (including ηk) inthe lowered position in this subsection, in contrast to the convention adopted in subsections 3.2 and 4.3.
25We will soon be suppressing the indices, so the reason for the bar on mij is merely to distinguish the lowered-index mass matrix.
42
βα
ji
p
ip·σαβ Cij
α β
i j
p
ip·σαβ (CT)ij
α β
i j
iδαβ Dij
α β
i j
iδαβ Dij
Figure 19: The full, loop-corrected propagators for two-component fermions are associatedwith functions C(p2)i
j [and its matrix transpose], D(p2)ij , and D(p2)ij , as shown. The shadedboxes represent the sum of all connected Feynman diagrams, with external legs included. Thefour-momentum p flows from right to left.
fully interacting two-component fermion fields [cf. footnotefnft]. Following eqs. (4.1)–(4.4), we
define:
〈0| Tψαi(x)ψjβ(y) |0〉FT = ip·σαβ Cij(p2) , (4.42)
〈0| T ψαi(x)ψβj (y) |0〉FT = ip·σαβ (CT) ij(p2) , (4.43)
〈0| T ψαi(x)ψjβ(y) |0〉FT = iδαβ Dij(p2) , (4.44)
〈0| Tψαi(x)ψβj (y) |0〉FT = iδαβ Dij(p
2) , (4.45)
where
(CT)ij ≡ Cji . (4.46)
One can derive eq. (4.43) from eq. (4.42) by first writing
ψαi(x)ψβj (y) = −εβα εαβψαj(y)ψiβ(x) , (4.47)
where the minus sign arises due to the anticommutativity of the fields, and then using eq. (2.22);
the interchange of x and y (after FT) simply changes pµ to −pµ.In general, D and D are complex symmetric matrices, and D = D?. The matrix C
satisfies the hermiticity condition [CT]? = C. Here, we have introduced the star symbol to
mean that a quantity Q? is obtained from Q by taking the complex conjugate of all Lagrangian
parameters appearing in its calculation, but not taking the complex conjugates of Euclideanized
loop integral functions, whose imaginary (absorptive) parts correspond to fermion decay widths
to multi-particle intermediate states. That is, the dispersive part of C is hermitian and the
absorptive part of C is anti-hermitian.
The diagrammatic representations of the full propagators are displayed in Fig. 19, where
Cij, Dij , and Dij defined above are each N×N matrix functions. Note that the second diagram
of Fig. 19, when flipped by 180 about the vertical axis, is equivalent to the first diagram of
Fig. 19 (with p→ −p, α→ β, β → α and i↔ j). In analogy with Fig. 3, one could replace the
first two diagrammatic rules of Fig. 19 with a single rule:
43
βα
ji
p
ip·σαβ Cij or −ip·σβα (CT)j i
where we have used eq. (4.46) to rewrite the second version of the rule in terms of CT. Indeed,
by using the second version of the above rule and flipping the corresponding diagram by 180
as described above, one reproduces the rule of the second diagram of Fig. 19.
In what follows, we prefer to keep the first two rules of Fig. 19 as separate entities. This
will permit us to conveniently assemble the four diagrams of Fig. 19 into a 2 × 2 block matrix
of two-component propagators [c.f. eq. (E.58)]. In addition, by choosing the momentum flow in
the two-component propagators from right to left, the left-to-right orderings of the spinor labels
of the diagrams coincide with the ordering of spinor indices that occurs in the corresponding
algebraic representations. Thus, we can multiply diagrams together and interpret them as the
product of the respective algebraic quantities taken from left to right in the normal fashion.
Starting at tree-level and comparing with Fig. 2, the full propagator functions are given by:
Cij = δi
j/(p2 −m2i ) + . . . (4.48)
Dij = mij/(p2 −m2i ) + . . . (4.49)
Dij = mij/(p2 −m2
i ) + . . . , (4.50)
with no sum on i in each case. They are functions of the external momentum invariant p2
and of the masses and couplings of the theory. Inserting the leading terms [eqs. (4.48)–(4.50)]
into Fig. 19 and organizing the result in a 2 × 2 block matrix of two-component propagators
reproduces the usual four-component fermion tree-level propagator given in eq. (E.58).
The computation of the full propagators can be organized, as usual in quantum field theory,
in terms of one-particle irreducible (1PI) self-energy functions. These are formally defined to be
the sum of all Feynman diagrams (excluding the tree-level) that contribute to the 1PI two-point
Green function. Diagrammatically, the 1PI self-energy functions are defined in Fig. 20.
p
βα
i j
−ip·σαβΞij
p
βα
i j
−ip·σαβ(ΞT)ij
α β
i j
−iδαβΩij
α β
i j
−iδαβΩij
Figure 20: The self-energy functions for two-component fermions are associated with functionsΞ(p2)i
j [and its matrix transpose], Ω(p2)ij , and Ω(p2)ij , as shown. The shaded circles representthe sum of all one-particle irreducible, connected Feynman diagrams, and the external legs areamputated. The four-momentum p flows from right to left.
44
As in the case of the full loop-corrected propagators, [ΞT]? = Ξ and Ω = Ω?, where the star
symbol was defined in the paragraph following eq. (4.47), and (ΞT)ij ≡ Ξji.
We illustrate the computation of the full propagator by considering first the following dia-
grammatic identity (with momentum p flowing from right to left):
βα
ji=
βα
ji
γα
ki
δ
`
β
j
γα
ki
δ
`
β
j
γα
ki
δ
`
β
j
γα
ki
δ
`
β
j
=
+ +
+ +
(4.51)
Similar diagrammmatic identities can be constructed for the three other full loop-corrected
propagators of Fig. 19. The resulting four equations can be neatly summarized by the following
matrix diagrammatic identity:
=
1 0
0 1
+
(4.52)
We have chosen the labeling and momentum flow in Figs. 19 and 20 such that the spinor
and flavor labels of the diagrams appear in the appropriate left-to-right order to permit the
interpretation of eq. (4.52) as a matrix equation. The algebraic representation of eq. (4.52) can
be written as F = T + TSF , where F is the matrix of full loop-corrected propagators, T is the
matrix of tree-level propagators and S is the matrix of self-energy functions. Multiplying26 on
the left by T−1 and on the right by F−1 yields T−1 = F−1 + S. Thus, F = [T−1 − S]−1. In
pictures:
=
−1
−
−1
. (4.53)
26Alternatively, one can solve eq. (4.52) by iteration. This yields:
F = T + TS(T + TS(T + TS(· · · ))) = T + TST + TSTST + . . . = T [1 + ST + (ST )2 + . . .] = T [1 − ST ]−1 ,
where in the last step, we have summed the geometric series. Taking the inverse of F = T [1−ST ]−1, multiplyingout the resulting expression and then taking the inverse of both sides yields eq. (4.53).
45
We evaluate the tree-level propagator matrix and its inverse using eqs. (4.48)–(4.50), keeping
in mind that the direction of momentum flow is from right to left:
=
1
p2 −m2i
(imij δα
β ip·σαβ δij
ip·σαβ δij imij δαβ
), (4.54)
−1
=
(imij δα
β −ip·σαβ δij−ip·σαβ δij imij δ
αβ
), (4.55)
where we follow the index structure defined in Figs. 19 and 20. Inserting eq. (4.55) into eq. (4.53),
one obtains a 4N × 4N matrix equation for the full propagator functions:
(iD ip·σC
ip·σCT iD
)=
(i(m + Ω) −ip·σ (1−ΞT)
−ip·σ (1−Ξ) i(m + Ω)
)−1
, (4.56)
where 1 is the N × N identity matrix. The right hand side of eq. (4.56) can be evaluated by
employing the following identity for the inverse of a block-partitioned matrix [71]:
(P QR S
)−1
=
((P −QS−1R)−1 (R− SQ−1P )−1
(Q− PR−1S)−1 (S −RP−1Q)−1
), (4.57)
under the assumption that all inverses appearing in eq. (4.57) exist. Applying this result to
eq. (4.56), we obtain
C−1 = p2(1−Ξ)− (m + Ω)(1 −ΞT)−1(m + Ω) , (4.58)
D−1 = p2(1−Ξ)(m + Ω)−1(1−ΞT)− (m + Ω) , (4.59)
D−1
= p2(1−ΞT)(m + Ω)−1(1−Ξ)− (m + Ω) . (4.60)
Note that eq. (4.60) is consistent with eq. (4.59) as Ξ? = ΞT.
The pole mass can be found most easily by considering the rest frame of the (off-shell)
fermion, in which the space components of pµ vanish. This reduces the spinor-index dependence
to a triviality. Setting pµ = (√s ; 0), we search for values of s where the inverse of the full
propagator has a zero eigenvalue. This is equivalent to setting the determinant of the inverse of
the full propagator to zero. Here we shall use the well-known formula for the determinant of a
block-partitioned matrix [71]:
det
(P Q
R S
)= det P det (S −RP−1Q) . (4.61)
The end result is that the poles of the full propagator (which are in general complex),
spole,j ≡M2j − iΓjMj , (4.62)
46
are formally the solutions to the non-linear equation27
det[s1− (1−ΞT)−1(m + Ω)(1−Ξ)−1(m + Ω)
]= 0 , (4.63)
with s ≡ p2.
Some care is required in using eq. (4.63), since the pole squared mass always has a non-
positive imaginary part, while the loop integrals used to find the self-energy functions are complex
functions of a real variable s that is given an infinitesimal positive imaginary part. Therefore,
eq. (4.63) should be solved iteratively by first expanding the self-energy function matrices Ξ,
Ω and Ω in a series in s about either m2j + iε or M 2
j + iε. The complex pole mass quantities
spole,j are renormalization-group and gauge invariant physical observables. Examples are given
in subsections 6.23 and 6.24.
The results of this section can be applied to an arbitrary collection of fermions (both
Majorana or Dirac). However, it is convenient to treat separately the case where all fermions
are Dirac fermions (consisting of pairs of two-component fields χi and ηi). As discussed in
Section 3.2, the Dirac fermion mass-eigenstates are defined in eq. (3.80) and are determined by
the singular value decomposition of the Dirac fermion mass matrix. With respect to the mass
basis, we denote the diagonal Dirac fermion mass matrix by M ij . The elements of this matrix
are real and non-negative. Nevertheless, it will be convenient as before to define M ij ≡M ij to
maintain covariance when manipulating tensors with flavor indices.
At tree-level, there are four propagators for each pair of χ and η fields as shown in Fig. 5.
The corresponding full, loop-corrected propagators are shown in Fig. 21.
χ χβα
ji
p
ip·σαβ SR ij
η ηα β
i j
p
ip·σαβ (ST
L)ij
η χα β
i j
iδαβ SDij
χ ηα β
i j
iδαβ (S T
D)ij
Figure 21: The full, loop-corrected propagators for Dirac fermions, represented by pairs oftwo-component (oppositely charged) fermion fields χi and ηi, are associated with functionsSR(p2)i
j , ST
L(p2)ij, SD(p2)ij , and S T
D(p2)ij , as shown. The shaded boxes represent the sum ofall connected Feynman diagrams, with external legs included. The four-momentum p and thecharge of χ flow from right to left.
The naming and sign conventions employed for the full, loop-corrected Dirac fermion propagator
functions in Fig. 21 derives from the correponding functions used in the more traditional four-
component treatment presented in Appendix E [c.f. eq. (E.79)].
27The determinant of the inverse of the full propagator [the inverse of eq. (4.56)] is equal to eq. (4.63) multipliedby det [−(1 − Ξ)(1 − Ξ
T)]. We assume that the latter does not vanish. This must be true perturbatively sincethe eigenvalues of Ξ are one-loop (or higher) quantities, which one assumes cannot be as large as 1.
47
In general, the complex matrices SR and SL satisfy hermiticity conditions [ST
R]? = SR and
[ST
L]? = SL, whereas the complex matrices SD and SD are related by SD = S ?D, where the
star symbol is defined in the paragraph below eq. (4.47). In contrast to the general case treated
earlier, SR and SL are unrelated and SD is a complex matrix (not necessarily symmetric).
Instead of working in a χ–η basis for the two-component Dirac fermion fields, one can
Takagi-digonalize the fermion mass matrix. In the new ψ-basis, the loop-corrected propagators
of Fig. 19 are applicable. It is easy to check that the number of independent functions is the
same in both methods for treating Dirac fermions. In particular, the loop-corrected propagator
functions in the ψ-basis are given in terms of the corresponding functions in the χ–η basis by:28
C =
(SR 00 SL
), D =
(0 ST
D
SD 0
), D =
(0 S
T
D
SD 0
). (4.64)
We similarly introduce the 1PI self-energy matrix functions for the Dirac fermions in the
χ–η basis, where the corresponding self-energy functions are defined in Fig. 22.
p
χ χβα
i j
−ip·σαβΣL ij
η η
p
βα
i j
−ip·σαβ(ΣT
R)ij
η χα β
i j
−iδαβΣDij
χ ηα β
i j
−iδαβ(ΣT
D)ij
Figure 22: The self-energy functions for two-component Dirac fermions, represented by pairsof two-component (oppositely charged) fermion fields χi and ηi, are associated with functionsΣL(p2)i
j , ΣT
R(p2)ij, ΣD(p2)ij, and ΣT
D(p2)ij , as shown. The shaded circles represent the sum ofall one-particle irreducible, connected Feynman diagrams, and the external legs are amputated.The four-momentum p flows from right to left.
As before, the naming and sign conventions employed for the Dirac fermion self-energy functions
above derives from the correponding functions used in the more traditional four-component
treatment of Appendix E [c.f. eq. (E.80)].
Once again, the complex matrices ΣL and ΣR satisfy hermiticity conditions [ΣT
L]? = ΣL
and [ΣT
R]? = ΣR, whereas the complex matrices ΣD and ΣD are related by ΣD = Σ?D, where
the star symbol is defined in the paragraph below eq. (4.47). Likewise, ΣL and ΣR are unrelated
and ΣD is a complex matrix (not necessarily symmetric). The self-energy functions in the ψ-
basis are given in terms of the corresponding functions in the χ–η basis by:28
Ξ =
(ΣL 00 ΣR
), Ω =
(0 ΣT
D
ΣD 0
), Ω =
(0 Σ
T
D
ΣD 0
). (4.65)
28The simple forms of C in eq. (4.64) and Ξ in eq. (4.65) motivate our definitions of SL and ΣR with thetranspose as indicated in Figs. 21 and 22, respectively.
48
In the case of Dirac fermions fields, eq. (4.53) still holds in the χ–η basis, which yields:
(iS T
D ip·σSR
ip·σ ST
L iSD
)=
(i(M + ΣD) −ip·σ (1−ΣT
R)
−ip·σ (1−ΣL) i(M + ΣT
D)
)−1
, (4.66)
Using eq. (4.57), it follows that:
SL−1 = p2(1−ΣR)− (M + ΣD)(1−ΣT
L)−1(M + ΣT
D) , (4.67)
SR−1 = p2(1−ΣL)− (M + Σ
T
D)(1−ΣT
R)−1(M + ΣD) , (4.68)
SD−1 = p2(1−ΣL)(M + ΣD)−1(1−ΣT
R)− (M + ΣT
D) , (4.69)
SD
−1= p2(1−ΣT
L)(M + ΣD)−1(1−ΣR)− (M + ΣT
D) . (4.70)
Note that eq. (4.70) is consistent with eq. (4.69) as Σ?L,R = ΣT
L,R.
The pole mass is now easily computed using the technique previously outlined. In particular,
eq. (4.63) is replaced by:
det[s1− (1−ΣT
R)−1(M + ΣD)(1−ΣL)−1(M + ΣT
D)]
= 0 , (4.71)
which determines the complex pole squared masses spole of the corresponding Dirac fermions.
Again, the self-energy functions should be expanded in a series in s about a point with an
infinitesimal positive imaginary part.
Finally, we examine the special case of a parity-conserving vectorlike theory of Dirac
fermions (such as QED or QCD). In this case, the following relations hold among the loop-
corrected propagator functions and self-energy functions, respectively:29
SRij = (ST
L)ij , SDij = (S T
D)ij , (4.72)
ΣLij = (ΣT
R)ij , ΣDij = (ΣT
D)ij . (4.73)
By imposing eq. (4.73) on eqs. (4.67)–(4.70) and recalling that M ij = M ij, it is straightforward
to verify that eq. (4.72) is satisfied.
5 Conventions for fermion and anti-fermion names and fields
In this section, we discuss conventions for labeling Feynman diagrams that contain two-component
fermion fields of the Standard Model (SM) and its minimal supersymmetric extension (MSSM).
In the case of Majorana fermions, there is a one-to-one correspondence between the par-
ticle names and the unbarred ( 12 , 0) [left-handed] fields. In contrast, for Dirac fermions there
are always two distinct two-component fields that correspond to each particle name. This is
illustrated in Table 1, which lists the SM and MSSM fermion particle names together with the
29These relations are derived using four-component spinor methods in Appendix E [cf. eqs. (E.87) and (E.88)].
49
Table 1: Fermion and anti-fermion names and two-component fields in the Standard Model andthe MSSM. In the listing of two-component fields, the first is an unbarred ( 1
2 , 0) [left-handed]field and the second is a barred (0, 1
2) [right-handed] field. (In this table, neutrinos are consideredto be exactly massless and the left-handed antineutrino ν c is absent from the spectrum).
Fermion name Two-component fields
`− (lepton) ` , `c
`+ (anti-lepton) `c , `
ν (neutrino) ν , −
ν (antineutrino) − , ν
q (quark) q , qc
q (anti-quark) qc , q
f (quark or lepton) f , f c
f (anti-quark or anti-lepton) f c , f
Ni (neutralino) χ0i , χ
0i
C+i (chargino) χ+
i , χ−i
C−i (anti-chargino) χ−
i , χ+i
g (gluino) g , g
corresponding two-component fields. For each particle, we list the two-component field with the
same quantum numbers, i.e., the field that contains a creation operator for that one-particle
state when acting to the right on the vacuum state |0〉.There is an option of labeling fermion lines in Feynman diagrams by particle names or by
field names; each choice has advantages and disadvantages.30 In all of the examples that follow,
we have chosen to eliminate the possibility of ambiguity as follows. We always label fermion lines
with two-component fields (rather than particle names), and adopt the following conventions:
• In the Feynman rules for interaction vertices, the external lines are always labeled by the
unbarred ( 12 , 0) [left-handed] field, regardless of whether the corresponding arrow is pointed in
or out of the vertex. Two-component fermion lines with arrows pointing away from the vertex
correspond to dotted indices, and two-component fermion lines with arrows pointing toward the
vertex always correspond to undotted indices. This also applies to Feynman diagrams where
the initial state and the final state roles are ambiguous (such as self-energy diagrams).
30Unfortunately, the notation for fermion names can be ambiguous because some of the symbols used alsoappear as names for one of the two-component fermion fields. In practice, it should be clear from the contextwhich set of names are being employed.
50
• Internal fermion lines in Feynman diagrams are also always labeled by the unbarred ( 12 , 0)
[left-handed] field(s). Internal fermion lines containing a propagator with opposing arrows can
carry two labels (see e.g. Fig. 15).
• Initial-state external fermion lines (which always have physical four-momenta pointing
into the vertex) in Feynman diagrams are labeled by the corresponding unbarred ( 12 , 0) [left-
handed] field if the arrow is into the vertex, and by the barred (0, 12) [right-handed] field if the
arrow is away from the vertex.
• Final-state external fermion lines in complete Feynman diagrams (which always have
physical four-momenta pointing out of the vertex) are labeled by the corresponding barred
(0, 12) [right-handed] field if the arrow is into the vertex, and by the unbarred ( 1
2 , 0) [left-handed]
field if the arrow is away from the vertex.
In particular, the field labels used for external fermion lines always correspond to the same
conserved quantities (charges, lepton numbers, baryon numbers) as the corresponding physical
particle. As an example, for either initial or final states, the two-component fields e and ec
both represent the negatively charged electron, conventionally denoted by e−, whereas both ec
and e represent the positively charged positron, conventionally denoted by e+ (as indicated in
Table 1). The rules for using these states as external particles are summarized in Fig. 23.
Initial-state e−:e ec
Initial-state e+:ec e
Final-state e−:e ec
Final-state e+:ec e
Figure 23: The two-component field labeling conventions for external fermion lines in a Feyn-man diagram for a physical process. The top row corresponds to an initial-state electron, thesecond row to an initial-state positron, the third row to a final-state electron, and the fourthrow to a final-state positron. The labels above each line are the two-component field names.The corresponding conventions for a massless neutrino are obtained by deleting the diagramswith ec or ec, and of course changing e and e to ν and ν.
51
Initial-state Ni:χ0i χ0
i
Final-state Ni:χ0i χ0
i
Figure 24: The two-component field labeling conventions for external neutralino lines in aFeynman diagram for a physical process. The top row corresponds to an initial-state neutralino,and the second row to a final-state neutralino. The labels above each line are the two-componentfield names. (The neutralino is its own antiparticle.)
The applicaiton of our naming conventions to processes involving Majorana fermions is
completely straightforward. For example, the conventions for employing the neutralino states
as external particles are summarized in Fig. 24.
As a simple example, consider Bhabha scattering (e−e+ → e−e+) [72]. We require the
the two-component Feynman rules for the QED coupling of electrons and positrons to the
photon, which are exhibited in Fig. 25. Consider the s-channel tree-level Feynman diagrams
that contribute to the invariant amplitude for e−e+ → e−e+. If we were to label the external
fermion lines according to the corresponding particle names (which does not conform to the
conventions introduced above), the result is shown in Fig. 26. One can find the identity of the
external two-component fermion fields by carefully observing the direction of the arrow of each
γ
α
β
ieσαβµ or −ieσµβα
e
e
(a)
γβ
α
−ieσαβµ or ieσµβα
ec
ec
(b)
Figure 25: The two-component Feynman rules for the QED vertex. Following the conventionsoutlined in this Section, we label these rules with the ( 1
2 , 0) [left-handed] fields e and ec, whichcomprise the Dirac electron. Note that e > 0 and Qe = −1 [cf. Fig. 67].
52
fermion line. For contrast, the same diagrams, relabeled with two-component fields following
the conventions established in this section (c.f. Fig. 23), are shown in Fig. 27. An explicit
computation of the invariant amplitude is given in Section 6.3.
e−
e+
e−
e+
e−
e+
e−
e+
e−
e+
e−
e+
e−
e+
e−
e+
Figure 26: Tree-level s-channel Feynman diagrams for e−e+ → e−e+, with the external lineslabeled according to the particle names. The initial state is on the left, and the final state is onthe right. Thus, the physical momentum flow of the external particles, as well as the flow of thelabeled charges, are indicated by the arrows adjacent to the corresponding fermion lines in theupper left diagram.
e
e
ec
ec
ec
ec
ec
ec
e
e
e
e
ec
ec
e
e
Figure 27: Tree-level s-channel Feynman diagrams for e+e− → e+e−. These diagrams are thesame as in Fig. 26, but with the external lines relabeled by the two-component fermion fieldsaccording to our conventions.
53
6 Practical examples from the Standard Model and supersym-metry
In this section we will present some examples to illustrate the use of the rules presented in this
paper. These examples are chosen from the Standard Model [73] and the MSSM [47–49], in
order to provide an unambiguous point of reference. In all cases, the fermion lines in Feynman
diagrams are labeled by two-component field names, rather than the particle names, as explained
in Section 5.
6.1 Top quark decay: t→ bW +
We begin by calculating the decay width of a top quark into a bottom quark and W + vector
boson. For simplicity, we treat this as a one-generation problem and ignore Cabibbo-Kobayashi-
Maskawa (CKM) [74] mixing among the three quark generations [see eq. (H.7) and the sur-
rounding text]. Let the four-momenta and helicities of these particle be (pt, λt), (kb, λb) and
(kW , λW ), respectively. Then p2t = m2
t , k2b = m2
b and k2W
= m2W
and
2pt ·kW = m2t −m2
b +m2W , (6.1)
2pt ·kb = m2t +m2
b −m2W , (6.2)
2kW ·kb = m2t −m2
b −m2W . (6.3)
Because only left-handed top quarks couple to the W boson, the only Feynman diagram for
t → bW+ is the one shown in Fig. 28. The corresponding amplitude can be read off of the
Feynman rule of Fig. 67 in Appendix H. Here the initial-state top quark is a two-component
field t going into the vertex and the final-state bottom quark is created by a two-component
field b. Therefore the amplitude is given by:
iM = −i g√2ε∗µxbσ
µxt , (6.4)
where ε∗µ ≡ ε∗µ(kW , λW ) is the polarization vector of the W+, and xb ≡ x(~kb, λb) and xt ≡x(~pt, λt) are the external-state wavefunction factors for the bottom and top quark. Squaring
this amplitude yields:
|M|2 =g2
2ε∗µεν(xbσ
µxt) (xtσνxb) , (6.5)
t(pt, λt)
W+(kW , λW )
b(kb, λb)
Figure 28: The Feynman diagram for t→ bW+ at tree level.
54
where we have used equation (2.32). Next, we can average over the top quark spin polarizations
using eq. (3.56):1
2
∑
λt
|M|2 =g2
4ε∗µεν xbσ
µ pt ·σ σνxb . (6.6)
Summing over the bottom quark spin polarizations in the same way yields a trace over spinor
indices:
1
2
∑
λt,λb
|M|2 =g2
4ε∗µεν Tr[σµpt ·σ σνkb ·σ] =
g2
2ε∗µεν
(pµt k
νb + kµb p
νt − gµνpt ·kb
), (6.7)
where we have used eq. (2.45). Finally we can sum over the W + polarizations according to:
∑
λW
ε∗µεν = −gµν + (kW )µ(kW )ν/m2W. (6.8)
The result is:
1
2
∑
spins
|M|2 =g2
2
[pt ·kb + 2(pt ·kW )(kb ·kW )/m2
W
]. (6.9)
After performing the phase space integration, one obtains:
Γ(t→ bW+) =1
16πm3t
λ1/2(m2t ,m
2W,m2
b)
1
2
∑
spins
|M|2 (6.10)
=g2
64πm2Wm
3t
λ1/2(m2t ,m
2W,m2
b)[(m2
t + 2m2W )(m2
t −m2W ) +m2
b(m2W − 2m2
t ) +m4b
], (6.11)
where
λ(x, y, z) ≡ x2 + y2 + z2 − 2xy − 2xz − 2yz. (6.12)
In the approximation mb mW ,mt, one ends up with the well-known result [75]
Γ(t→ bW+) =g2mt
64π
(2 +
m2t
m2W
)(1−
m2W
m2t
)2
, (6.13)
which exhibits the Nambu-Goldstone enhancement factor (m2t /m
2W
) for the longitudinal W
contribution compared to the two transverse W contributions [75].
6.2 Z0 vector boson decay: Z0 → f f
Consider the partial decay width of the Z0 boson into Standard Model fermion-antifermion final
states. There are two Feynman diagrams (as in the generic example of Fig. 12), shown in Fig. 29.
In the first diagram, the fermion particle f in the final state is created by a two-component field
f in the Feynman rule, and the anti-fermion particle f by a two-component field f . In the
second diagram, the fermion particle f in the final state is created by a two-component field
55
Z0(p, λZ)
f(kf , λf )
f(kf , λf )
Z0(p, εµ)
f c(kf , λf )
f c(kf , λf )(a) (b)
Figure 29: The Feynman diagrams for Z0 decay into a fermion-antifermion pair. Fermion linesare labeled according to the two-component fermion field labeling convention (see Section 5).
f c, and the anti-fermion particle f by a two-component field f c. Let us call the initial Z0 four-
momentum and helicity (p, λZ) and the final state fermion (f) and anti-fermion (f) momentum
and helicities (kf , λf ) and (kf , λf ), respectively. Then, k2f = k2
f= m2
f and p2 = m2Z, and
kf ·kf =1
2m2Z−m2
f , (6.14)
p·kf = p·kf = 12m
2Z. (6.15)
According to the rules of Fig. 67, the matrix elements for the two Feynman graphs are:
iMa = −i gcW
(T f3 −Qfs2W ) εµxfσµyf , (6.16)
iMb = igQf
s2W
cWεµyfσ
µxf , (6.17)
where xi ≡ x(~ki, λi) and yi ≡ y(~ki, λi), for i = f, f , and εµ ≡ εµ(p, λZ).
Using the Bouchiat-Michel formulae developed in Appendix F, one can explicitly evaluate
Ma andMb as a function of the final state fermion helicities. The result of this computation is
given in eqs. (F.64) and (F.65). If the final state helicities are not measured, then it is simpler
to square the amplitude and sum over the final state spins.
It is convenient to define:
af ≡ T f3 −Qfs2W ; bf ≡ −Qfs2W. (6.18)
Then the squared matrix element for the decay is, using eqs. (2.31) and (2.32),
|M|2 =g2
c2W
εµε∗ν
(af xfσ
µyf + bfyfσµxf
)(af yfσ
νxf + bfxfσν yf
). (6.19)
Summing over the anti-fermion helicity using eqs. (3.56)–(3.59) gives:
∑
λf
|M|2 =g2
c2W
εµε∗ν
(a2f xfσ
µkf ·σσνxf + b2fyfσµkf ·σσν yf
−mfafbf xfσµσν yf −mfaf bfyfσ
µσνxf
). (6.20)
56
Next, we sum over the fermion helicity:
∑
λf ,λf
|M|2 =g2
c2W
εµε∗ν
(a2fTr[σµkf ·σσνkf ·σ] + b2fTr[σµkf ·σσνkf ·σ]
−m2fafbfTr[σµσν ]−m2
faf bfTr[σµσν ]). (6.21)
Averaging over the Z0 polarization using
1
3
∑
λZ
εµεν∗ =1
3
(−gµν + pµpν/m
2Z
)(6.22)
and applying eqs. (2.43)–(2.45), one gets:
1
3
∑
spins
|M|2 =g2
3c2W
[(a2f + b2f )
(2kf ·kf + 4 kf ·p kf ·p/m2
Z
)+ 12af bfm
2f
](6.23)
=2g2
3c2W
[(a2f + b2f )(m
2Z−m2
f ) + 6af bfm2f
], (6.24)
where we have used eqs. (6.14) and (6.15). After the standard phase-space integration, we obtain
the well-known result for the partial width of the Z 0:
Γ(Z0 → f f) =Nfc
16πmZ
(1−
4m2f
m2Z
)1/21
3
∑
spins
|M|2 (6.25)
=Nfc g2mZ
24πc2W
(1−
4m2f
m2Z
)1/2 [(a2f + b2f )
(1−
m2f
m2Z
)+ 6af bf
m2f
m2Z
]. (6.26)
Here we have also included a factor of N fc (equal to 1 for leptons and 3 for quarks) for the sum
over colors. (Since the Z0 is a color singlet, the color factor is simply equal to the dimension of
the color representation of the outgoing fermions.)
6.3 Bhabha scattering: e−e+ → e−e+
In our next example, we consider the computation of Bhabha scattering in QED (that is, we
we consider photon exchange but neglect Z0-exchange) [72]. Bhabha scattering has also been
computed using two-component spinors in [62]. We denote the initial state electron and positron
momenta and helicities by (p1, λ1) and (p2, λ2) and the final state electron and positron momenta
and helicities by (p3, λ3) and (p4, λ4), respectively. Neglecting the electron mass, we have in
terms of the usual Mandelstam variables s, t, u:
p1 ·p2 = p3 ·p4 ≡ 12s , (6.27)
p1 ·p3 = p2 ·p4 ≡ −12 t , (6.28)
p1 ·p4 = p2 ·p3 ≡ −12u , (6.29)
57
ec
e
ec
e
e
e
e
e
ec
ec
ec
ec
e
ec
e
ec
Figure 30: Tree-level t-channel Feynman diagrams for e−e+ → e−e+, with the external lineslabeled according to the two-component field names. The momentum flow of the externalparticles is from left to right.
and p2i = 0 for i = 1, . . . , 4. There are eight distinct Feynman diagrams. First, there are four
s-channel diagrams, as shown in Fig. 27 with amplitudes that follow from the Feynman rules of
Fig. 25 (more generally, see Fig. 67 in Appendix H):
iMs =
(−igµνs
)[(−ie x1σµy2)(ie y3σν x4) + (−ie y1σµx2)(ie y3σν x4)
+(−ie x1σµy2)(ie x3σνy4) + (−ie y1σµx2)(ie x3σνy4)], (6.30)
where xi ≡ x(~pi, λi) and yi ≡ y(~pi, λi), for i = 1, 4. The photon propagator in Feynman gauge
is −igµν/(p1 + p2)2 = −igµν/s. Here, we have chosen to follow the fermion lines in the order
1, 2, 3, 4. This dictates in each term the use of either the σ or σ forms of the Feynman rules of
Fig. 25. One can group the terms of eq. (6.30) together more compactly:
iMs = e2(−igµν
s
)(x1σµy2 + y1σµx2) (y3σν x4 + x3σνy4) . (6.31)
There are also four t-channel diagrams, as shown in Fig. 30. The corresponding amplitudes
for these four diagrams can be written:
iMt = (−1)e2(−igµν
t
)(x1σµx3 + y1σµy3) (x2σν x4 + y2σνy4) . (6.32)
Here, the overall factor of (−1) comes from Fermi-Dirac statistics, since the external fermion
wave functions are written in an odd permutation (1, 3, 2, 4) of the original order (1, 2, 3, 4)
established by the first term in eq. (6.30).
Fierzing each term using eqs. (2.55)–(2.57), and using eqs. (2.47) and (2.48), the total
58
amplitude can be written as:
M =Ms +Mt = 2e2[1
s(x1y3)(y2x4) +
1
s(y1x3)(x2y4) +
(1
s+
1
t
)(y1x4)(x2y3)
+
(1
s+
1
t
)(x1y4)(y2x3)−
1
t(x1x2)(x3x4) −
1
t(y1y2)(y3y4)
]. (6.33)
Squaring this amplitude and summing over spins, all of the cross-terms will vanish in the me → 0
limit. This is because each cross term will have an x or an x for some electron or positron
combined with a y or a y for the same particle, and the corresponding spin sum is proportional
tome [see eqs. (3.58) and (3.59)]. Hence, summing over final state spins and averaging over initial
state spins, the end result contains only the sum of the squares of the six terms in eq. (6.33):
1
4
∑
spins
|M|2 = e4∑
λ1,λ2,λ3,λ4
1
s2[(x1y3)(y3x1)(y2x4)(x4y2) + (y1x3)(x3y1)(x2y4)(y4x2)]
+
(1
s+
1
t
)2
[(y1x4)(x4y1)(x2y3)(y3x2) + (x1y4)(y4x1)(y2x3)(x3y2)]
+1
t2[(x1x2)(x2x1)(x3x4)(x4x3) + (y1y2)(y2y1)(y3y4)(y4y3)]
. (6.34)
Here we have used eq. (2.30) to get the complex square of the fermion bilinears. Performing
these spin sums using eqs. (3.56) and (3.57) and using the trace identities eq. (A.3):
1
4
∑
spins
|M|2 = 8e4[p2 ·p4 p1 ·p3
s2+p1 ·p2 p3 ·p4
t2+
(1
s+
1
t
)2
p1 ·p4 p2 ·p3
]
= 2e4[t2
s2+s2
t2+(us
+u
t
)2]. (6.35)
Thus, the differential cross section for Bhabha scattering is given by:
dσ
dt=
1
16πs2
1
4
∑
spins
|M|2 =
2πα2
s2
[t2
s2+s2
t2+(us
+u
t
)2]. (6.36)
This agrees with the result of, for example, problem 5.2 of ref. [76].
6.4 Polarized Muon Decay
So far we have only treated cases where the initial state fermion spins are averaged and the final
state spins are summed. In the case of the polarized decay of a particle or polarized scattering
we must project out the appropriate polarization of the particles in the spin sums. This is
achieved by replacing the spin sums given in eqs. (3.56)-(3.59) by the corresponding polarized
spin projections eqs. (3.31)-(3.34). As an example, we consider the decay of a polarized muon.
Polarized muon decay has also been computed using two-component spinors in [62], however
with an effective four-fermion interaction.
59
µ (p, s) e (ke, λe)
νe (kνe , λνe)
νµ (kνµ , λνµ)
W−
Figure 31: Feynman diagram for electroweak muon decay.
In Fig. 31, we show the single leading order Feynman diagram for muon decay, including
the definition of the momenta. We denote the mass of the muon by mµ, and neglect the electron
mass. We shall assume the muon is polarized with a (contravariant) spin four-vector31
sρ = (0; 0, 0, 1) , (6.37)
along the z-axis in the muon rest frame. The amplitude is then given by
iM =
(−ig√2
)2 (xνµσρxµ
)(xeστyνe)
(−igρτDW
), (6.38)
where DW = (p − kνµ)2 − M2
W is the denominator of the W -boson propagator. Here xµ ≡x(~p, s) for the spin-polarized initial state muon, and xνµ ≡ x(~kνµ , λνµ), xe ≡ x(~ke, λe), and
yνe ≡ y(~kνe , λνe). Squaring the amplitude using eq. (2.32), we obtain
|M|2 =g4
4D2W
(xνµσ
ρxµ) (xµσ
τxνµ)(xeσρyνe) (yνeστxe) (6.39)
Summing over the neutrino and electron spins using eqs. (3.56)-(3.57), and using eq. (3.44) for
the muon spin yields:
∑
λνµλeλνe
|M|2 =g4
8D2W
Tr[kνµ ·σ σρ(p·σ −mµs·σ)στ ] Tr[ke ·σ σρkνe ·σ στ ] , (6.40)
=2g4
D2W
ke ·kνµ kνe ·(p−mµs). (6.41)
To obtain the second line we have used the trace identity eq. (2.44) twice; note that the resulting
terms linear in the antisymmetric tensor do not contribute, but the term quadratic in the
antisymmetric tensor does.
The differential decay amplitude is now given by
dΓ =1
2mµ|M|2 d3~ke
(2π)32Ee
d3~kνe(2π)32Eνe
d3~kνµ
(2π)32Eνµ(2π)4δ4(p− ke − kνe − kνµ) , (6.42)
31Throughout this subsection µ and ν are particle labels and not Lorentz vector indices. Instead we use ρ, τ .
60
φ0
f (p1, λ1)
f c (p2, λ2)
(a)
φ0
f c (p2, λ2)
f (p1, λ1)
(b)
Figure 32: The Feynman diagrams for the decays φ0 → f f , where φ0 = h0,H0, A0 are theneutral Higgs scalar bosons of minimal supersymmetry, and f is a Standard Model quark orlepton, and f is the corresponding antiparticle. We have labeled the external fermions accordingto the two-component field names.
where Ei, i = e, νe, νµ are the energies of the final state particles in the muon rest frame. In
the following we shall neglect both the electron mass and the momentum in the W -propagator
compared to the W -boson mass, so DW → −M2W . We can now use the following identity to
integrate over the neutrino momenta [77]
∫d3~kνe
(2π)32Eνe
d3~kνµ
(2π)32Eνµ(2π)4δ4(q − kνe − kνµ)kρνekτνµ =
1
96π(q2gρτ + 2qρqτ ) , (6.43)
where q = p− ke. It follows that
dΓ =g4
1536π4mµM4W
[q2 ke ·(p−mµs) + 2q ·ke q ·(p−mµs)
] d3~ke
Ee(6.44)
In the muon rest frame ke = Ee(1; cos φ sin θ, sinφ sin θ, cos θ), so that q2 = m2µ − 2Eemµ and
ke ·(p−mµs) = mµEe(1 + cos θ) and q ·ke = mµEe and q ·(p−mµs) = mµ(mµ −Ee −Ee cos θ).
Noting that the maximum energy of the electron is mµ/2 (when the neutrino and antineutrino
both recoil in the opposite direction), we obtain
dΓ
d(cos θ)=
g4m2µ
768π3M4W
∫ mµ/2
0dEeE
2e
[3− 4Ee
mµ+
(1− 4Ee
mµ
)cos θ
](6.45)
=g4m5
µ
12288π3M4W
(1− 1
3cos θ
)(6.46)
in agreement with ref. [77].
6.5 Neutral Higgs boson decays φ0 → ff , for φ0= h0, H0, A0 in supersymmetry
In this subsection, we consider the decays of the neutral Higgs scalar bosons φ0 = h0, H0, and A0
of minimal supersymmetry into Standard Model fermion-antifermion pairs. The relevant tree-
level Feynman diagrams are shown in Fig. 32. The final state fermion is assigned four-momentum
p1 and polarization λ1, and the antifermion is assigned four-momentum p2 and polarization λ2.
We will first work out the case that f is a charge −1/3 quark or a charged lepton, and later
61
note the simple change needed for charge +2/3 quarks. Then the Feynman rules of Figure 70
of Appendix I tell us that the resulting amplitudes are:
iMa =i√2Yf k
∗dφ0 y1y2 , (6.47)
iMb =i√2Yf kdφ0 x1x2 . (6.48)
Here Yf is the Yukawa coupling of the fermion, kdφ0 is the Higgs mixing parameter from eq. (I.6),
and the external wave functions are denoted x1 ≡ x(~p1, λ1), y1 ≡ y(~p1, λ1) for the fermion
and x2 ≡ x(~p2, λ2), y2 ≡ y(~p2, λ2) for the antifermion. Squaring the total amplitude iM =
iMa + iMb using eq. (2.30) results in:
|M|2 =1
2|Yf |2
[|kdφ0 |2(y1y2 y2y1 + x1x2 x2x1) + (kdφ0)2x1x2 y2y1 + (k∗dφ0)
2y1y2 x2x1
]. (6.49)
Summing over the final-state antifermion spin using eqs. (3.56)-(3.59) gives:
∑
λ2
|M|2 =1
2|Yf |2
[|kdφ0 |2(y1p2 ·σy1 + x1p2 ·σx1)− (kdφ0)2mf x1y1 − (k∗dφ0)
2mfy1x1
]. (6.50)
Summing over the fermion spins in the same way yields:
∑
λ1,λ2
|M|2 =1
2|Yf |2
|kdφ0 |2(Tr[p2 ·σp1 ·σ] + Tr[p2 ·σp1 ·σ])− 2(kdφ0)2m2
f − 2(k∗dφ0)2m2
f
(6.51)
= |Yf |22|kdφ0 |2p1 ·p2 − 2Re[(kdφ0)2]m2
f
(6.52)
= |Yf |2|kdφ0 |2(m2
φ0 − 2m2f )− 2Re[(kdφ0)2]m2
f
, (6.53)
where we have used the trace identity eq. (2.43) to obtain the second equality. The corresponding
expression for charge +2/3 quarks can be obtained by simply replacing kdφ0 with kuφ0 . The total
decay rates now follow from integration over phase space [78]
Γ(φ0 → ff) =Nfc
16πmφ0
(1− 4m2
f/m2φ0
)1/2 ∑
λ1,λ2
|M|2. (6.54)
The factor of N fc = 3 for quarks and 1 for leptons comes from the sum over colors.
Results for special cases are obtained by putting in the relevant values for the couplings
and the mixing parameters including eqs. (I.5) and (I.6). In particular, for the CP-even Higgs
bosons h0 and H0, kdφ0 and kuφ0 are real, so one obtains:
Γ(h0 → bb) =3
16πY 2b sin2 αmh0
(1− 4m2
b/m2h0
)3/2, (6.55)
Γ(h0 → cc) =3
16πY 2c cos2 αmh0
(1− 4m2
c/m2h0
)3/2, (6.56)
Γ(h0 → τ+τ−) =1
16πY 2τ sin2 αmh0
(1− 4m2
τ/m2h0
)3/2, (6.57)
Γ(H0 → tt) =3
16πY 2t sin2 αmH0
(1− 4m2
t /m2H0
)3/2, (6.58)
Γ(H0 → bb) =3
16πY 2b cos2 αmH0
(1− 4m2
b/m2H0
)3/2, (6.59)
62
νe
e (ke, λe)
χ+i (kC , λC)
Figure 33: The Feynman diagram for νe → C+i e
− in the MSSM.
etc., which check with the expressions in Appendix B of ref. [79]. For the pseudo-scalar Higgs
boson A0, the mixing parameters kuA0 = i cos β0 and kdA0 = i sin β0 are purely imaginary, so
Γ(A0 → tt) =3
16πY 2t cos2 β0mA0
(1− 4m2
t /m2A0
)1/2, (6.60)
Γ(A0 → bb) =3
16πY 2b sin2 β0mA0
(1− 4m2
b/m2A0
)1/2, (6.61)
Γ(A0 → τ+τ−) =1
16πY 2τ sin2 β0mA0
(1− 4m2
τ/m2A0
)1/2. (6.62)
Note that the differing kinematic factors for the pseudo-scalar decays came about because
of the different relative sign between the two Feynman diagrams. For example, in the case of
h0 → bb, the matrix element is
iM =i√2Yb sinα(y1y2 + x1x2), (6.63)
while for A0 → bb, it is
iM =1√2Yb sinβ0(y1y2 − x1x2). (6.64)
The differing sign follows from the imaginary pseudo-scalar Lagrangian coupling, which is com-
plex conjugated in the second diagram.
6.6 Sneutrino decay νe → C+i e−
Next we consider the process of sneutrino decay νe → C+i e
− in minimal supersymmetry. Be-
cause only the left-handed electron can couple to the chargino and sneutrino (with the excellent
approximation that the electron Yukawa coupling is 0), there is just one Feynman diagram,
shown in Fig. 33. The external wave functions of the electron and chargino are denoted as
xe ≡ x(~ke, λe), and xC ≡ x(~kC , λC), respectively. From the corresponding Feynman rule given
in Fig. 75 of Appendix I, the amplitude is:
iM = −igVi1 xC xe, (6.65)
where Vij is one of the two matrices used to diagonalize the chargino masses [cf. eq. (I.21].
Squaring this using eq. (2.30) yields:
|M|2 = g2|Vi1|2 (xC xe)(xexC) . (6.66)
63
χ+i (pC , λC)
e (ke, λe)
νe
Figure 34: The Feynman diagram for C+i → νee
+ in the MSSM.
Now summing over the electron and chargino spin polarizations using eq. (3.56) yields
∑
λe,λC
|M|2 = g2|Vi1|2Tr[ke ·σ kC ·σ] = 2g2|Vi1|2 ke ·kC = g2|Vi1|2(m2νe −m2
Ci) , (6.67)
where we have used 2ke ·kC = m2νe−m2
Ci, neglecting the electron mass. Therefore, after inte-
grating over phase space in the standard way, the decay width is:
Γ(νe → C+i e
−) =1
16πmνe
(1−
m2Ci
m2νe
)∑
λe,λC
|M|2 =
g2
16π|Vi1|2mνe
(1−
m2Ci
m2νe
)2
, (6.68)
which agrees with ref. [80] and eq. (3.8) in ref. [48].
6.7 Chargino Decay C+i → νee
+
Here again, there is just one Feynman diagram (neglecting the electron mass in the Yukawa
coupling) shown in Fig. 34. The external wave functions for the chargino and the positron
are denoted by xC ≡ x(~pC , λC) and ye ≡ y(~ke, λe), respectively. The fermion momenta and
helicities are denoted as in Fig. 34. As in the previous example, the amplitude can be directly
determined using the Feynman rule given in Fig. 75 in Appendix I:
M = −igV ∗i1 xC ye . (6.69)
Squaring this using eq. (2.30) yields:
|M|2 = g2|Vi1|2 (xCye) (yexC) . (6.70)
Summing over the electron helicity and averaging over the chargino helicity using eqs. (3.56)
and (3.57) we obtain:
12
∑
λe,λC
|M|2 = 12g
2|Vi1|2Tr[ke ·σ pC ·σ] = g2|Vi1|2ke ·pC =g2
2|Vi1|2(m2
Ci−m2
νe) . (6.71)
So the decay width is, neglecting the electron mass:
Γ(C+i → νe+) =
1
16πmCi
(1−
m2νe
m2Ci
)1
2
∑
λe,λC
|M|2 =
g2
32π|Vi1|2mCi
(1−
m2νe
m2Ci
)2
, (6.72)
which agrees with ref. [80].
64
χ0i (pi, λi)
χ0j (kj , λj)
φ0
χ0i (pi, λi)
χ0j (kj , λj)
φ0
Figure 35: The Feynman diagrams for Ni → Njφ0 in the MSSM.
6.8 Neutralino Decays Ni → φ0Nj, for φ0= h0, H0, A0
Next we consider the decay of a neutralino to a lighter neutralino and neutral Higgs scalar
boson φ0 = h0, H0, or A0. The two tree-level Feynman graphs are shown in Fig. 35, where we
have also labeled the momenta and helicities. We denote the masses for the neutralinos and the
Higgs boson as m eNi, m eNj
, and mφ0 . Using the Feynman rules of Fig. 74, the amplitudes are
respectively given by
iM1 = −iY xiyj , (6.73)
iM2 = −iY ∗ yixj , (6.74)
where the coupling Y ≡ Y φ0χ0iχ
0j is defined in eq. (I.25), and the external wave functions are
xi ≡ x(~pi, λi), yi ≡ y(~pi, λi), yj ≡ y(~kj, λj), and xj ≡ x(~kj , λj).Taking the square of the total matrix element using eq. (2.30) gives:
|M|2 = |Y |2(xiyj yjxi + yixjxjyi) + Y 2xiyjxjyi + Y ∗2yixj yjxi. (6.75)
Now summing over the final-state neutralino spins using eqs. (3.56)-(3.59) yields
∑
λj
|M|2 = |Y |2(xikj ·σxi + yikj ·σyi)− Y 2m eNjxiyi − Y ∗2m eNj
yixi. (6.76)
Averaging over the initial-state neutralino spins in the same way gives
1
2
∑
λi,λj
|M|2 =1
2|Y |2(Tr[kj ·σpi ·σ] + Tr[kj ·σpi ·σ]) + Re[Y 2]m eNi
m eNjTr[1] (6.77)
= 2|Y |2pi ·kj + 2Re[Y 2]m eNim eNj
(6.78)
= |Y |2(m2eNi
+m2eNj−m2
φ0) + 2Re[Y 2]m eNim eNj
, (6.79)
where we have used eq. (2.43) to obtain the second equality. The total decay rate is therefore
Γ(Ni → φ0Nj) =1
16πm3eNi
λ1/2(m2eNi,m2
φ0 ,m2eNj
)
1
2
∑
λi,λj
|M|2 (6.80)
=m eNi
16πλ1/2(1, rφ, rj)
[|Y φ0χ0
iχ0j |2(1 + rj − rφ) + 2Re
[(Y φ0χ0
iχ0j)2]√
rj
], (6.81)
65
χ0i (pi, λi)
χ0j (kj , λj)
Z0 (kZ , λZ)
χ0i (pi, λi)
χ0j (kj , λj)
Z0 (kZ , λZ)
Figure 36: The Feynman diagrams for Ni → NjZ0 in the MSSM.
where rj ≡ m2eNj/m2
eNiand rφ ≡ m2
φ0/m2eNi
. The results for φ0 = h0,H0, A0 can now be obtained
by using eqs. (I.5) and (I.6) in eq. (I.25). In comparing eq. (6.81) with the original calculation
in [81], it is helpful to employ eqs. (4.51) and (4.53) of [82]. The results agree.
6.9 Ni → Z0Nj
For this two-body decay there are two tree-level Feynman diagrams, shown in Fig. 36 with the
definitions of the helicities and the momenta. The two amplitudes are given by32
iM1 = −i gcWO′′Lji xiσ
µxjε∗µ (6.82)
iM2 = ig
cWO′′Lij yiσ
µyjε∗µ (6.83)
where the external wave functions are xi = x(~pi, λi), yi = y(~pi, λi), xj = x(~kj, λj), yj =
y(~kj, λj), and ε∗µ = ε∗µ(~kZ , λZ). Noting that O′′Lji = O′′L∗
ij [see eq. (I.20)], and applying eqs. (2.31)
and (2.32), we find that the matrix element squared is:
|M|2 =g2
c2Wε∗µεν
[|O′′L
ij |2(xiσµxjxjσν xi + yiσµyj yjσ
νyi) (6.84)
−(O′′Lij
)2yiσ
µyjxjσν xi −
(O′′L∗ij
)2xiσ
µxj yjσνyi
](6.85)
Now summing over the final-state neutralino spin using eqs. (3.56)-(3.59) yields:
∑
λj
|M|2 =g2
c2Wε∗µεν
[|O′′L
ij |2(xiσµkj ·σσν xi + yiσµkj ·σσνyi) (6.86)
+(O′′Lij
)2m eNj
yiσµσν xi +
(O′′L∗ij
)2m eNj
xiσµσνyi
]. (6.87)
Averaging over the initial-state neutralino spins in the same way gives
1
2
∑
λi,λj
|M|2 =g2
2c2Wε∗µεν
[|O′′L
ij |2(Tr[σµkj ·σσνpi ·σ] + Tr[σµkj ·σσνpi ·σ]
)(6.88)
−(O′′Lij
)2m eNi
m eNjTr[σµσν ]−
(O′′L∗ij
)2m eNi
m eNjTr[σµσν ]
](6.89)
=2g2
c2Wε∗µεν
|O′′L
ij |2(kµj p
νi + pµi k
νj − pi ·kjgµν
)−Re
[(O′′Lij
)2]m eNi
m eNjgµν, (6.90)
32When comparing with the 4-component Feynman rule in ref. [48] note that O′′Lij = −O′′R∗
ij , c.f. eq. (I.20).
66
e (p1, λ1)
ec (p2, λ2)
e−L (k1)
e−R (k2)
χ0i
e (p2, λ2)
ec p1, λ1) e−L (k1)
e−R (k2)
χ0i
Figure 37: Feynman diagrams for e−e− → e−L e−R.
where in the last equality we have applied eqs. (2.43)-(2.45). Now using
∑
λZ
εµ∗εν = −gµν + kµZkνZ/m
2Z , (6.91)
we obtain
1
2
∑
λi,λj ,λZ
|M|2 =2g2
c2W
|O′′L
ij |2(−pi ·kj + 2pi ·kZkj ·kZ/m2
Z
)+ 3m eNi
m eNjRe[(O′′Lij
)2]
(6.92)
and, noting that and 2kj ·kZ = m2eNi− m2
eNj− m2
Z , 2pi ·kj = m2eNi
+ m2eNj− m2
Z , and 2pi ·kZ =
m2eNi−m2
eNj+m2
Z , we get the total decay width
Γ(Ni → Z0Nj) =1
16πm3eNi
λ1/2(m2
eNi,m2
Z,m2
eNj
)1
2
∑
λi,λj ,λZ
|M|2 (6.93)
=g2m eNi
16πc2Wλ1/2(1, rZ , rj)
[|O′′L
ij |2(1 + rj − 2rZ + (1− rj)2/rZ
)+ 6Re
[(O′′Lij
)2]√rj
], (6.94)
where
rj ≡ m2eNj/m2
eNi, and rZ ≡ m2
Z/m2eNi, (6.95)
and the triangle kinematic function was defined in eq. (6.12). The result eq. (6.94) agrees with
the original calculation in [81].
6.10 Selectron pair production in electron-electron collisions
6.10.1 e−e− → e−L e−R
Here there are two Feynman graphs (neglecting the electron mass and Yukawa couplings), shown
in Fig. 37. Note that these two graphs are related by interchange of the identical initial state
electrons. Let the electrons have momenta p1 and p2 and the selectrons have momenta keeL and
keeR , so that p21 = p2
2 = 0; k21 = m2
eL; k2
2 = m2eR
; s = (p1 + p2)2 = (k1 + k2)
2; t = (k1 − p1)2 =
(k2−p2)2; u = (k1−p2)
2 = (k2−p1)2. The matrix element for the first graph, for each neutralino
Ni exchanged in the t channel, is:
iMt =
[ig√2
(N∗i2 +
sWcW
N∗i1
)][−i√
2gsWcW
Ni1
]x1
[i(k1 − p1)·σ
(k1 − p1)2 −m2Ni
]y2 . (6.96)
67
Here we have used the Feynman rules from Fig. 76. We employ the notation for the external
wave functions xi = (~pi, λi), i = 1, 2 and analogously for yi, xi, yi. The matrix elements for the
second (u-channel) graph are the same with the two incoming electrons exchanged, e1 ↔ e2:
iMu = (−1)
[ig√2
(N∗i2 +
sWcW
N∗i1
)][−i√
2gsWcW
Ni1
]x2
[i(k1 − p2)·σ
(k1 − p2)2 −m2Ni
]y1 . (6.97)
Note that since we have written the fermion wave function spinors in the opposite order inM2
compared toM1, there is a factor (−1) for Fermi-Dirac statistics. Alternatively, starting at the
electron with momentum p1 and using the Feynman rules as above, we can directly write:
iMu =
[ig√2
(N∗i2 +
sWcW
N∗i1
)][−i√
2gsWcW
Ni1
]y1
[−i(k1 − p2)·σ
(k1 − p2)2 −m2Ni
]x2 . (6.98)
This has no Fermi-Dirac factor (−1) because the wave function spinors are written in the same
order as inMt. However, now the Feynman rule for the propagator has an extra minus sign, as
can be seen in Fig. 3. We can also obtain eq. (6.98) from eq. (6.97) by using eq. (2.49). So we
can write for the total amplitude:
M =Mt +Mu = x1a·σy2 + y1b·σx2 , (6.99)
where
aµ ≡ g2sWcW
(kµ1 − pµ1 )
4∑
i=1
Ni1(N∗i2 +
sWcW
N∗i1)
1
t−m2Ni
, (6.100)
bµ ≡ −g2sWcW
(kµ1 − pµ2 )
4∑
i=1
Ni1(N∗i2 +
sWcW
N∗i1)
1
u−m2Ni
. (6.101)
So, using eqs. (2.31) and (2.32):
|M|2 = (x1a·σy2) (y2a∗ ·σx1) + (y1b·σx2) (x2b
∗ ·σy1) + (x1a·σy2) (x2b∗ ·σy1)
+ (y1b·σx2) (y2a∗ ·σx1) . (6.102)
Averaging over the initial state electron spins using eqs. (3.56)-(3.59), the a, b∗ and a∗, b cross
terms are proportional to me and can thus be neglected in our approximation. We get:
1
4
∑
λ1,λ2
|M|2 =1
4Tr[a·σ p2 ·σ a∗ ·σ p1 ·σ] +
1
4Tr[b·σ p2 ·σ b∗ ·σ p1 ·σ] . (6.103)
These terms can be simplified using the identities:
Tr[(k1 − p1)·σ p2 ·σ (k1 − p1)·σ p1 ·σ] = Tr[(k1 − p2)·σ p2 ·σ (k1 − p2)·σ p1 ·σ] (6.104)
= tu−m2eLm
2eR , (6.105)
68
ec (p1, λ1)
ec (p2, λ2)
e−R (k1)
e−R (k2)
χ0i
ec (p2, λ2)
ec (p1, λ1) e−R (k1)
e−R (k2)
χ0i
Figure 38: The two Feynman diagrams for e−e− → e−R e−R in the limit where me → 0.
which follow from eq. (2.44) and (2.45), resulting in:
1
4
∑
λ1,λ2
|M|2 =g4s2W4c2W
(tu−m2eLm
2eR)
4∑
i,j=1
Nj1N∗i1(N
∗j2 +
sWcW
N∗j1)(Ni2 +
sWcW
Ni1)
[1
(t−m2Ni
)(t−m2Nj
)+
1
(u−m2Ni
)(u−m2Nj
)
]. (6.106)
To get the differential cross-section dσ/dt, multiply this by 1/(16πs2):
dσ
dt=
πα2
4s2W c2W
(tu−m2
eeLm2
eeR
s2
)4∑
i,j=1
Nj1N∗i1(N
∗j2 +
sWcW
N∗j1)(Ni2 +
sWcW
Ni1)
[1
(t−m2Ni
)(t−m2Nj
)+
1
(u−m2Ni
)(u−m2Nj
)
]. (6.107)
To compare with the original calculation in [83] and with eq. E26, p. 244 in ref. [48], note that
for a pure photino exchange, Ni1 → cW δi1 and Ni2 → sW δi1, so that
1
4s2W c2W
|Ni1|2|Ni2 +sWcW
Ni1|2 → 1 . (6.108)
Also note that in [83] polarized electron beams are assumed. The result checks.
6.10.2 e−e− → e−Re−R
For this process, there are again two Feynman graphs, which are related by the exchange of
identical electrons in the initial state or equivalently by exchange of the identical selectrons in
the final state, as shown in Fig. 38. (We again neglect the electron mass and thus the Higgsino
coupling to the electron.) Let the electrons have momenta p1 and p2 and the selectrons have
momenta k1 and k2, so that p21 = p2
2 = 0; k21 = k2
2 = m2eeR
; s = (p1 + p2)2; t = (k1 − p1)
2;
u = (k1 − p2)2. Using the Feynman rules of Fig. 76, the amplitude for the first graph is:
iMt =
(−i√
2gsWcW
Ni1
)2[
im eNi
(k1 − p1)2 −m2eNi
]y1y2 (6.109)
69
for each exchanged neutralino. The amplitudes for the second graph are the same, but with the
electrons interchanged:
iMu =
(−i√
2gsWcW
Ni1
)2[
im eNi
(k1 − p2)2 −m2eNi
]y1y2 . (6.110)
Since we have chosen to write the external state wave function spinors in the same order inMt
andMu, there is no factor of (−1) for Fermi-Dirac statistics. So, applying eq. (2.30), the total
amplitude squared is:
|M|2 =4g4s4Wc4W
(y1y2)(y2y1)
4∑
i,j=1
(Ni1)2(N∗
j1)2m eNi
m eNj
(1
t−m2eNi
+1
u−m2eNi
) 1
t−m2eNj
+1
u−m2eNj
(6.111)
The sum over the electron spins is obtained from
∑
λ1,λ2
(y1y2)(y2y1) = Tr[p2 ·σp1 ·σ] = 2p2 ·p1 = s . (6.112)
So, using eq. (3.57), the spin-averaged differential cross-section is:
dσ
dt=
(1
2
)1
16πs2
1
4
∑
λ1,λ2
|M|2 (6.113)
=πα2
2c4W
4∑
i,j=1
(Ni1)2(N∗
j1)2m eNi
m eNj
s
(1
t−m2eNi
+1
u−m2eNi
) 1
t−m2eNj
+1
u−m2eNj
.(6.114)
The first factor of (1/2) in eq. (6.113) comes from the fact that there are identical sleptons in
the final state and thus the phase space is degenerate.
To compare with [83] and also with eq. E27, p. 245 in ref. [48], note that for a pure photino
exchange, Ni1 → cW δi1, so it checks.
6.10.3 e−e− → e−L e−L
Again, in the limit of vanishing electron mass, there are two Feynman graphs, which are related
by the exchange of identical electrons in the initial state or equivalently by exchange of the
identical selectrons in the final state. As shown in Fig. 39, they are exactly like the previous
example, but with all arrows reversed. Using the Feynman rules of Fig. 76, the amplitude for
the first graph is:
Mt =
(ig√2[N∗
i2 +sWcW
N∗i1]
)2[
im eNi
(p1 − k1)2 −m2eNi
]x1x2 (6.115)
70
e (p1, λ1)
e (p2, λ2)
e−L (k1)
e−L (k2)
χ0i
e (p2, λ2)
e (p1, λ1) e−L (k1)
e−L (k2)
χ0i
Figure 39: The two Feynman diagrams for e−e− → e−L e−L in the limit of vanishing electron
mass.
for each exchanged neutralino. The amplitudes for the second graph are the same, but with
p1 ↔ p2:
Mu =
(ig√2[N∗
i2 +sWcW
N∗i1]
)2[
im eNi
(p2 − k1)2 −m2eNi
]x1x2 (6.116)
Since we have chosen to write the external state wave function spinors in the same order inM1
andM2, there is no factor of (−1) for Fermi-Dirac statistics. The total amplitude squared is:
|M|2 =g4
4(x1x2)(x2x1)
4∑
i,j=1
(N∗i2 +
sWcW
N∗i1)
2(Nj2 +sWcW
Nj1)2m eNi
m eNj
(1
t−m2eNi
+1
u−m2eNi
) 1
t−m2eNj
+1
u−m2eNj
. (6.117)
The average over the electron spins follows from eq. (3.56):
∑
λ1,λ2
(x1x2)(x2x1) = Tr[p2 ·σp1 ·σ] = 2p2 ·p1 = s . (6.118)
So the spin-averaged differential cross-section is:
dσ
dt=
(1
2
)1
16πs2
1
4
∑
λ1,λ2
|M|2 (6.119)
=πα2
32s4W
4∑
i,j=1
(N∗i2 +
sWcW
N∗i1)
2(Nj2 +sWcW
Nj1)2m eNi
m eNj
s(
1
t−m2eNi
+1
u−m2eNi
) 1
t−m2eNj
+1
u−m2eNj
(6.120)
where the first factor of (1/2) in eq. (6.119) comes from the fact that there are identical sleptons
in the final state. To compare with [83] and also with eq. (E27), p. 245 in ref. [48], note that for
a pure photino exchange, Ni1 → cW δi1 and Ni2 → sW δi1, so it checks.
71
e(p1, λ1)
e(p2, λ2)
ν(k1)
ν∗(k2)
Z0
ec(p1, λ1)
ec(p2, λ2)
ν(k1)
ν∗(k2)
Z0
e (p1, λ1)
e (p2, λ2)
ν (k1)
ν∗ (k2)
χ+i
Figure 40: The Feynman diagrams for e−e+ → νν∗.
6.11 e−e+ → νν∗
Consider now the pair-production of sneutrinos in electron-positron collisions. There are two
graphs featuring the s-channel exchange of the Z 0. We will neglect the electron mass and Yukawa
coupling, so there is only one graph involving the t-channel exchange of the charginos. These
three Feynman diagrams are shown in in Fig. 40, where we have also defined the helicities and
momenta of the particles. The Mandelstam variables can be expressed in terms of the external
momenta and the sneutrino mass:
2p1 ·p2 = s ; 2k1 ·k2 = s− 2m2ν ; (6.121)
2p1 ·k1 = 2p2 ·k2 = m2ν − t ; (6.122)
2p1 ·k2 = 2p2 ·k1 = m2ν − u . (6.123)
Using the Feynman rules of Fig. 67, the amplitudes for the two s-channel Z boson exchange
diagrams are:
iM1 =
[−i g
2cW(k1 − k2)µ
] [ −igµν(p1 + p2)2 −m2
Z + iΓZmZ
] [ig
cW(s2W − 1/2)
]x1σν y2 (6.124)
iM2 =
[−i g
2cW(k1 − k2)µ
] [ −igµν(p1 + p2)2 −m2
Z + iΓZmZ
] [igs2WcW
]y1σνx2 (6.125)
where the first factor in each case is the Feynman rule from the Z boson coupling to the sneutrinos
(see Fig. 72c, ref. [48]). The t-channel diagram due to each chargino gives a contribution
iM3 = (−igV ∗i1) (−igVi1) x1
[i(k1 − p1)·σ
(k1 − p1)2 −m2Ci
]y2, (6.126)
72
using the rules of Fig. 75. Therefore, the total amplitude can be rewritten as:
M = c1x1(k1 − k2)·σy2 + c2y1(k1 − k2)·σx2 + c3x1(k1 − p1)·σy2 (6.127)
where
c1 ≡g2(1− 2s2W )
4c2WDZ, c2 ≡ −
g2s2W2c2WDZ
, c3 ≡ g22∑
i=1
|Vi1|2/(m2Cj− t). (6.128)
and DZ ≡ s−m2Z + iΓZmZ is the denominator of the Z boson propagator.
We will now square the amplitude and sum over the electron and positron spins. In doing
so, the interference terms involving c2 will vanish for me = 0, because of eqs. (3.58) and (3.59).
Therefore, we have
∑
λ1,λ2
|M|2 =∑
λ1,λ2
|c1|2 x1(k1 − k2)·σy2 y2(k1 − k2)·σx1
+|c2|2 y1(k1 − k2)·σx2 x2(k1 − k2)·σy1
+c23 x1(k1 − p1)·σy2 y2(k1 − p1)·σx1
+2Re[c1c3 x1(k1 − k2)·σy2 y2(k1 − p1)·σx1]
(6.129)
= |c1|2 Tr[(k1 − k2)·σp2 ·σ(k1 − k2)·σp1 ·σ]
+|c2|2 Tr[(k1 − k2)·σp2 ·σ(k1 − k2)·σp1 ·σ]
+c23 Tr[(k1 − p1)·σp2 ·σ(k1 − p1)·σp1 ·σ]
+2Re[c1]c3 Tr[(k1 − k2)·σp2 ·σ(k1 − p1)·σp1 ·σ], (6.130)
where we have used eqs. (3.56) and (3.57) to do the spin sums to obtain the second equality. Now,
applying the trace identities (2.44) and (2.45) and reducing the results using eqs. (6.121)-(6.123)
and u = 2m2ν − s− t, we get
∑
λ1,λ2
|M|2 = −[st+ (t−m2ν)
2](4|c1|2 + 4|c2|2 + c23 + 4Re[c1]c3
). (6.131)
When mC1= mC2
, this agrees with eqs. (E46)-(E48) of ref. [48]33 and with [84]. The differential
cross-section follows in the standard way by averaging over the initial-state spins:
dσ
dt=
1
16πs2
(1
4
∑
λ1,λ2
|M|2). (6.132)
Note that
t = m2ν − (1− β cos θ)s/2; β ≡ (1− 4m2
ν/s)1/2, (6.133)
33There is a typo in eq. (E48) of [48]; the right-hand side should be multiplied by 1/cos2 θw.
73
where θ is the angle between the initial-state electron and the final-state sneutrino in the center-
of-momentum frame. The upper and lower limits t+ and t− are obtained by inserting cos θ = ±1
above, respectively.
Doing the integration over t to obtain the total cross-section, one obtains
σ =
∫ t+
t−
dσ
dtdt =
g4
64πs
(SZ +
2∑
i,j=1
Sij +
2∑
i=1
SZi
), (6.134)
where
SZ =β3
24c4W(8s4W − 4s2W + 1)
s2
|DZ |2, (6.135)
Sii = |Vi1|4 [(1− 2γi)Li − 2β] , (6.136)
S12 = S21 = |V11V12|2[(m2
C2+ sγ2
2)L2 − (m2C1
+ sγ21)L1]/(m
2C2−m2
C1)− β
, (6.137)
SZi =(2s2W − 1)
c2W|Vi1|2
[(m2
Ci+ sγ2
i )Li + sβ(γi − 1/2)] (s−m2
Z)
|DZ |2, (6.138)
with
γi ≡ (m2ν −m2
Ci)/s, Li ≡ ln
(m2Ci− t−
m2Ci− t+
). (6.139)
This agrees with eqs. (E49)-(E52) of ref. [48] in the limit of degenerate charginos, or of a single
wino chargino with |V11| = 1 and V12 = 0 and with [84].
6.12 e−e+ → NiNj
Next we consider the pair production of neutralinos via e−e+ annihilation. There are four
Feynman graphs for s-channel Z0 exchange, shown in Figure 41, and four for t-channel selectron
exchange, shown in Figure 42. The momenta and polarizations are as labeled in the graphs.
We denote the neutralino masses as m eNi,m eNj
and the selectron masses as meL and meR . The
electron mass will again be neglected. The kinematic variables are then given by
s = 2p1 ·p2 = m2eNi
+m2eNj
+ 2ki ·kj , (6.140)
t = m2eNi− 2p1 ·ki = m2
eNj− 2p2 ·kj , (6.141)
u = m2eNi− 2p2 ·ki = m2
eNj− 2p1 ·kj . (6.142)
By applying the Feynman rules of figs. 67 and 72, we obtain for the sum of the s-channel
diagrams in Fig. 41,
iMZ =−igµνs−m2
Z
[ig(s2W − 1
2)
cWx1σµy2 +
igs2WcW
y1σµx2
][ig
cWO′′Lij xiσνyj −
ig
cWO′′Lji yiσν xj
], (6.143)
where O′′ij is given in eq. (I.20). The fermion spinors are denoted by x1 ≡ x(~p1, λ1), y2 ≡
y(~p2, λ2), xi ≡ x(~ki, λi), yj ≡ y(~kj , λj), etc. Note that we have combined the matrix elements
74
of the four diagrams by factorizing with respect to the common boson propagator. For the four
t and u channel diagrams, we obtain, by applying the rules of Fig. 76:
iM(t)eL
= (−1)
[i
t−m2eL
][ ig√2
(N∗i2 +
sWcW
N∗i1
)][ ig√2
(Nj2 +
sWcW
Nj2
)]x1yiy2xj , (6.144)
iM(u)eL
=
[i
u−m2eL
][ ig√2
(N∗j2 +
sWcW
N∗j1
)][ ig√2
(Ni2 +
sWcW
Ni2
)]x1yj y2xi, (6.145)
iM(t)eR
= (−1)
[i
t−m2eR
](−i√
2gsWcW
Ni1
)(−i√
2gsWcW
N∗j1
)y1xix2yj , (6.146)
iM(u)eR
=
[i
u−m2eR
](−i√
2gsWcW
Nj1
)(−i√
2gsWcW
N∗i1
)y1xjx2yi. (6.147)
The first factors of (−1) in each of eqs. (6.144) and (6.146) are present because the order of the
spinors in each case is an odd permutation of the ordering (1, 2, i, j) established by the s-channel
contribution. The other contributions have spinors in an even permutation of that ordering.
The s-channel diagram contribution of eq. (6.143) can be profitably rearranged using the
Fierz identities of eqs. (2.55)-(2.56). Then, combining the result with the t-channel and s-channel
contributions, we have for the total:
M = c1x1yj y2xi + c2x1yiy2xj + c3y1xix2yj + c4y1xjx2yi, (6.148)
e (p1, λ1)
e (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
Z0
ec (p1, λ1)
ec (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
Z0
e (p1, λ1)
e (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
Z0
ec (p1, λ1)
ec (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
Z0
Figure 41: The four Feynman diagrams for e−e+ → NiNj via s-channel Z0 exchange.
75
e (p1, λ1)
e (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
eL
ec (p1, λ1)
ec (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
e∗R
e (p1, λ1)
e (p2, λ2)
χ0i (ki, λj)
χ0j (kj , λj)
eL
ec (p1, λ1)
ec (p2, λ2)
χ0i (ki, λi)
χ0j (kj , λj)
e∗R
Figure 42: The four Feynman diagrams for e−e+ → NiNj via t-channel selectron exchange.
where
c1 =g2
c2W
[(1− 2s2W )O′′L
ij /(s−m2Z)− (cWNi2 + sWNi1)(cWN
∗j2 + sWN
∗j1)/2(u −m2
eL)], (6.149)
c2 =g2
c2W
[(2s2W − 1)O′′L
ji /(s−m2Z) + (cWN
∗i2 + sWN
∗i1)(cWNj2 + sWNj1)/2(t−m2
eL)], (6.150)
c3 =2g2s2Wc2W
[−O′′L
ij /(s−m2Z) +Ni1N
∗j1/(t−m2
eR)], (6.151)
c4 =2g2s2Wc2W
[O′′Lji /(s−m2
Z)−N∗i1Nj1/(u−m2
eR)]. (6.152)
Now, when we square the amplitude and average over the initial-state fermion spins, the only
terms that will survive in the massless electron limit are the ones that involve x1x1 or y1y1, and
x2x2 or y2y2. This follows immediately from eqs. (3.58) and (3.59).
∑
λ1,λ2
|M|2 =∑
λ1,λ2
(|c1|2yjx1x1yjxiy2y2xi + |c2|2yix1x1yixjy2y2xj
+|c3|2xiy1y1xiyj x2x2yj + |c4|2xjy1y1xj yix2x2yi
+2Re[c1c
∗2yix1x1yjxjy2y2xi
]+ 2Re
[c3c
∗4xjy1y1xiyix2x2yj
])(6.153)
= |c1|2yjp1 ·σyj xip2 ·σxi + |c2|2yip1 ·σyi xjp2 ·σxj+|c3|2xip1 ·σxi yjp2 ·σyj + |c4|2xjp1 ·σxj yip2 ·σyi+2Re
[c1c
∗2yip1 ·σyj xjp2 ·σxi
]+ 2Re
[c3c
∗4xjp1 ·σxi yip2 ·σyj
](6.154)
where eqs. (3.56) and (3.57) have been used to do the spin sums to obtain the second equality.
76
Now we do the remaining spin sums using eqs. (3.56)-(3.59) again, obtaining:
∑
λ1,λ2,λi,λj
|M|2 = |c1|2Tr[p1 ·σkj ·σ]Tr[p2 ·σki ·σ] + |c2|2Tr[p1 ·σki ·σ]Tr[p2 ·σkj ·σ]
+|c3|2Tr[p1 ·σki ·σ]Tr[p2 ·σkj ·σ] + |c4|2Tr[p1 ·σkj ·σ]Tr[p2 ·σki ·σ]
+2Re[c1c∗2]m eNi
m eNjTr[p2 ·σp1 ·σ] + 2Re[c3c
∗4]m eNi
m eNjTr[p1 ·σp2 ·σ]. (6.155)
Applying the trace identity of eq. (2.43) to this yields
∑
spins
|M|2 = (|c1|2 + |c4|2)4p1 ·kj p2 ·ki + (|c2|2 + |c3|2)4p1 ·ki p2 ·kj+4Re[c1c
∗2 + c3c
∗4]m eNi
m eNjp1 ·p2 (6.156)
= (|c1|2 + |c4|2)(u−m2eNi
)(u−m2eNj
) + (|c2|2 + |c3|2)(t−m2eNi
)(t−m2eNj
)
+2Re[c1c∗2 + c3c
∗4]m eNi
m eNjs. (6.157)
The differential cross-section then follows:
dσ
dt=
1
16πs2
1
4
∑
spins
|M|2 . (6.158)
This agrees with the first complete calculation presented in [85]. For the case of pure photino
pair production, i.e. Ni1 → cW δi1 and Ni2 → sW δi1 and for degenerate selectron masses this also
agrees with eq. (E9) of the erratum of [48]. Other earlier calculations with some simplifications
are given in [86, 87].
Defining cos θ = p1 ·ki (the cosine of the angle between the initial-state electron and one of
the neutralinos in the center-of-momentum frame), the Mandelstam variables t, u can be written
as
t =1
2
[m2
eNi+m2
eNj− s+ λ1/2(s,m2
eNi,m2
eNj) cos θ
], (6.159)
u =1
2
[m2
eNi+m2
eNj− s− λ1/2(s,m2
eNi,m2
eNj) cos θ
]. (6.160)
Taking into account the identical fermions in the final state, the total cross section is
σ =1
2
∫ t+
t−
dσ
dtdt , (6.161)
where t− and t+ are obtained by inserting cos θ = ∓1 in eq. (6.159), respectively.
6.13 e−e+ → C−i C+
j
Next we consider the pair production of charginos in electron-positron collisions. The s-channel
Feynman diagrams are shown in Fig. 43, where we have also introduced the notation for the
77
e (p1, λ1)
e (p2, λ2)
χ−i (ki, λi)
χ−j (kj , λj)
γ, Z0
ec (p1, λ1)
ec (p2, λ2)
χ−i (ki, λi)
χ−j (kj , λj)
γ, Z0
e (p1, λ1)
e (p2, λ2)
χ+i (ki, λi)
χ+j (kj , λj)
γ, Z0
ec (p1, λ1)
ec (p2, λ2)
χ+i (ki, λi)
χ+j (kj , λj)
γ, Z0
Figure 43: Feynman diagrams for e−e+ → C−i C
+j via s-channel γ and Z0 exchange.
e (p1, λ1)
e (p2, λ2)
χ+i (ki, λi)
χ−j (kj , λj)
νe
Figure 44: The Feynman diagram for e−e+ → C−i C
+j via the t-channel exchange of a sneutrino.
fermion momenta and polarizations. The Mandelstam variables are given by
s = 2p1 ·p2 = m2eCi
+m2eCj
+ 2ki ·kj , (6.162)
t = m2eCi− 2p1 ·ki = m2
eCj− 2p2 ·kj , (6.163)
u = m2eCi− 2p2 ·ki = m2
eCj− 2p1 ·kj . (6.164)
Note that the negatively charged chargino carries momentum and polarization (ki, λi), while the
positively charged one carries (kj , λj). Using the Feynman rules of Figs. 67 and 72, the sum of
the photon-exchange diagrams is:
iMγ =−igµνs
(−ie x1σµy2 − ie y1σµx2) (ie δijyiσν xj + ie δij xiσνyj) (6.165)
while the result from the Z-exchange diagrams is:
iMZ =−igµνs−m2
Z
[ igcW
(s2W − 12)x1σµy2 +
igs2WcW
y1σµx2
][− ig
cWO′Lji yiσν xj −
ig
cWO′Rji xiσνyj
].(6.166)
78
The t-channel Feynman diagram via sneutrino exchange is shown in Fig. 44. Applying the rules
of Fig. 75, we find:
iMνe = (−1)i
t−m2νe
(−igV ∗i1x1yi) (−igVj1y2xj) . (6.167)
The Fermi-Dirac factor (−1) in this equation arises because the spinors appear an order which
is an odd permutation of the order used in all of the s-channel diagram results.
One can now apply the Fierz transformation identities (2.55)-(2.57) to eqs. (6.165) and
(6.166) to remove the σ and σ matrices. The result can be combined with the t-channel contri-
bution to obtain a total matrix element M with exactly the same form as eq. (6.148), but now
with:
c1 = 2e2δij/s−g2
c2W(1 − 2s2W )O′R
ji /(s−m2Z), (6.168)
c2 = 2e2δij/s−g2
c2W(1 − 2s2W )O′L
ji /(s−m2Z) + g2V ∗
i1Vj1/(t−m2νe), (6.169)
c3 = 2e2δij/s+2g2s2Wc2W
O′Rji /(s−m2
Z), (6.170)
c4 = 2e2δij/s+2g2s2Wc2W
O′Lji /(s−m2
Z). (6.171)
The rest of this calculation is identical in form to eqs. (6.148)-(6.157), so that the result is:∑
spins
|M|2 = (|c1|2 + |c4|2)(u−m2eCi
)(u−m2eCj
) + (|c2|2 + |c3|2)(t−m2eCi
)(t−m2eCj
)
+2Re[c1c∗2 + c3c
∗4]m eCi
m eCjs. (6.172)
The differential cross-section then follows:
dσ
dt=
1
16πs2
1
4
∑
spins
|M|2 . (6.173)
Defining cos θ = p1 ·ki (the cosine of the angle between the initial-state electron and C−i in the
center-of-momentum frame), the Mandelstam variables t, u can be written as
t =1
2
[m2
eCi+m2
eCj− s+ λ1/2(s,m2
eCi,m2
eCj) cos θ
], (6.174)
u =1
2
[m2
eCi+m2
eCj− s− λ1/2(s,m2
eCi,m2
eCj) cos θ
]. (6.175)
The total cross section can now be computed as
σ =
∫ t+
t−
dσ
dtdt (6.176)
where t− and t+ are obtained with cos θ = −1 and +1 in eq. (6.174), respectively. Our results
agree with the original first complete calculation in [88]. Earlier work with simplifying assump-
tions is given in [89]. An extended calculation for the production of polarized charginos is given
in [90].
79
u (p1, λ1)
d (p2, λ2)
χ+i (ki, λi)
χ0j (kj , λj)
W+
u (p1, λ1)
d (p2, λ2)
χ−i (ki, λi)
χ0j (kj , λj)
W+
u (p1, λ1)
d (p2, λ2)
χ−i (ki, λi)
χ0j (kj , λj)
dL
u (p1, λ1)
d (p2, λ2)
χ+i (ki, λi)
χ0j (kj , λj)
uL
Figure 45: The four tree-level Feynman diagrams for ud→ C+i Nj .
6.14 ud→ C+i Nj
Next we consider the associated production of a chargino and a neutralino in quark, anti-quark
collisions. The leading order Feynman diagrams are shown in Fig. 45, where we have also defined
the momenta and the helicities. The corresponding Mandelstam variables are
s = 2p1 ·p2 = m2eCi
+m2eNj
+ 2ki ·kj, (6.177)
t = m2eCi− 2p1 ·ki = m2
eNj− 2p2 ·kj, (6.178)
u = m2eCi− 2p2 ·ki = m2
eNj− 2p1 ·kj. (6.179)
The matrix elements for the s-channel diagrams are obtained by applying the Feynman
rules of figs. 67 and 73:
iMs =−igµνs−m2
W
(ig√2x1σµy2
)(igOL∗
ji xiσνyj + igOR∗ji yiσν xj
). (6.180)
The external spinors are denoted by x1 ≡ x(~p1, λ1), y2 ≡ y(~p2, λ2), xi ≡ x(~ki, λi), yj ≡ y(~kj, λj),etc. The matrix elements for the t and u channel graphs follow from the rules of figs. 75 and 76:
iMt = (−1)i
t−m2dL
(−igU∗i1)( ig√
2
[Nj2 −
sW3cW
Nj1
])x1yiy2xj (6.181)
iMu =i
u−m2uL
(−igVi1)( ig√
2
[−N∗
j2 −sW3cW
N∗j1
])x1yj y2xi (6.182)
The first factor of (−1) in eq. (6.181) is required because the order of the spinors (1, i, 2, j) is in
an odd permutation of the order (1, 2, i, j) used in the s-channel and u-channel results.
80
Now we can use the Fierz relations eqs. (2.55) and (2.57) to rewrite the s-channel amplitude
in a form without σ or σ matrices. Combining the result with the t-channel and u-channel
contributions yields a total M with exactly the same form as eq. (6.148), but now with
c1 = −√
2g2
[OL∗ji /(s−m2
W ) + Vi1
(1
2N∗j2 +
sW6cW
N∗j1)/(u−muL
)], (6.183)
c2 = −√
2g2
[OR∗ji /(s−m2
W ) + U∗i1
(1
2N∗j2 −
sW6cW
N∗j1)/(t−mdL
)], (6.184)
c3 = c4 = 0. (6.185)
The rest of this calculation is identical in form to that of eqs. (6.148)-(6.157), leading to:
∑
spins
|M|2 = |c1|2(u−m2eCi
)(u−m2eNj
) + |c2|2(t−m2eCi
)(t−m2eNj
) + 2Re[c1c∗2]m eCi
m eNjs. (6.186)
From this, one can obtain:
dσ
dt=
1
16πs2
1
3 · 4∑
spins
|M|2 , (6.187)
where we have included a factor of 1/3 from the color average for the incoming quarks. Eq. (6.187)
can be expressed in terms of the angle between the u quark and the chargino in the center-of-
momentum frame, using
t =1
2
[m2
eCi+m2
eNj− s+ λ1/2(s,m2
eCi,m2
eNj) cos θ
], (6.188)
u =1
2
[m2
eCi+m2
eNj− s− λ1/2(s,m2
eCi,m2
eNj) cos θ
]. (6.189)
This process occurs in proton-antiproton and proton-proton collisions, where√s is not fixed,
and the angle θ is different than the lab frame angle. The usable cross-section depends crucially
on experimental cuts. Our result in Eq. (6.187) agrees with the complete computation in [91].
Earlier calculations in special supersymmetric scenarios, e.g. with photino mass eigenstates are
given in [87, 92]
6.15 Ni → NjNkN`
Next we consider the decay of a neutralino Ni to three lighter neutralinos: Nj, Nk, N`. This
decay is not likely to be phenomenologically relevant, because a variety of two-body decay modes
will always be available. Furthermore, the calculation itself is quite complicated because of the
large number of Feynman diagrams involved. Therefore, we consider this only as a matter-
of-principle example of a process with four external-state Majorana fermions, and will restrict
ourselves to writing down the contributing matrix element amplitudes.
At tree-level, the decay can proceed via a virtual Z 0 boson; the Feynman graphs are shown
in Fig. 46. In addition, it can proceed via the exchange of any of the neutral scalar Higgs bosons
81
χ0i
χ0k
χ0l
χ0j
Z0
χ0i
χ0k
χ0l
χ0j
Z0
χ0i χ0
k
χ0l
χ0j
Z0
χ0i
χ0k
χ0l
χ0j
Z0
Figure 46: Four Feynman diagrams for Ni → NjNkN` in the MSSM via Z0 exchange. There
are four more where Nj ↔ Nk and another four where Nj ↔ N`.
χ0i
χ0k
χ0`
χ0j
h0,H0, A0
χ0i
χ0k
χ0`
χ0j
h0,H0, A0
χ0i
χ0k
χ0`
χ0j
h0,H0, A0
χ0i χ0
k
χ0`
χ0j
h0,H0, A0
Figure 47: Four Feynman diagrams for Ni → NjNkN` in the MSSM via φ0 = h0, H0, A0
exchange. There are four more where Nj ↔ Nk and another four where Nj ↔ N`.
of the MSSM, φ0 = h0,H0, A0, as shown in Fig. 47. Since any of the final state neutralinos can
directly couple to the initial state neutralino there are two more diagrams for each one shown
in Figs. 46 and 47, for a total of 48 tree-level diagrams (counting each intermediate Higgs boson
state as distinct). In all cases, the four-momenta of the neutralinos Ni, Nj , Nk, N` are denoted
pi, kj , kk, k` respectively.
82
For the sum of the four diagrams in Fig. 46, we obtain by implementing the rules of Fig. 72,
and using the Feynman gauge:
iM(1)Z =
−ig2/c2W(pi − kj)2 −m2
Z
(O′′Lji xiσµxj −O′′L
ij yiσµyj
)(O′′L`k xkσ
µy` −O′′Lk` ykσ
µx`
), (6.190)
[The external wave functions are xi ≡ x(~pi, λi), xj,k,` ≡ x(~kj,k,`, λj,k,`), and analogously for
xi,j,k,`, and yi,j,k,` and yi,j,k,`.] Note that we have factorized the sum of diagrams, taking advan-
tage of the common virtual boson line propagator. The contributions from the diagrams related
to these by permutations can now be obtained from the appropriate substitutions (j ↔ k) and
(j ↔ `):
iM(2)Z = (−1)
−ig2/c2W(pi − kk)2 −m2
Z
(O′′Lki xiσµxk −O′′L
ik yiσµyk
)(O′′L`j xjσ
µy` −O′′Lj` yjσ
µx`
), (6.191)
iM(3)Z = (−1)
−ig2/c2W(pi − k`)2 −m2
Z
(O′′L`i xiσµx` −O′′L
i` yiσµy`
)(O′′Ljk xkσ
µyj −O′′Lkj ykσ
µxj
). (6.192)
The first factors of (−1) in iM(2)Z and iM(3)
Z are present because the order of the spinors in each
case appear in an odd permutation of the canonical order set by iM(1)Z . Note that if we were
to proceed to a computation of the decay rate, the very first step would be to apply the Fierz
relations of eqs. (2.55)-(2.57) to eliminate all of the σ and σ matrices in the above amplitudes.
The diagrams in Fig. (47) combine to give a contribution:
iM(1)φ0 =
−i(pi − kj)2 −m2
φ0
(Y ijxiyj + Yij yixj)(Yk`yky` + Yk`xkx`) (6.193)
where we have adopted the shorthand notation Y ij = (Yij)∗ = Y φ0χ0
iχ0j . Again we have factored
the amplitude using the common virtual boson propagator. As in the Z-exchange diagrams, the
other contributions can be obtained by the appropriate substitutions:
iM(2)φ0 = (−1)
−i(pi − kk)2 −m2
φ0
(Y ikxiyk + Yikyixk)(Yj`yjy` + Yj`xj x`) (6.194)
iM(3)φ0 = (−1)
−i(pi − k`)2 −m2
φ0
(Y i`xiy` + Yi`yix`)(Ykjykyj + Ykjxkxj) (6.195)
The first factors of (−1) in iM(2)φ0 and iM(3)
φ0 are needed because the spinors in each case are in
an odd permutation of the canonical order established earlier.
Now the total matrix element is obtained by:
M =3∑
n=1
M(n)Z +
∑
φ0
3∑
n=1
M(n)φ0 (6.196)
as given above. When computing the total decay rate, additional attention must be paid to
the case where two or more final state indices are equal, since the phase space is then reduced
by the corresponding factor to avoid over-counting of identical final states. To the best of our
knowledge this process has not been computed in the literature.
83
˜−R (p) τ c (k2, λ2)
τ−1 (k3)
`c (k1, λ1)
χ0j
˜−R (p) τ (k2, λ2)
τ−1 (k3)
`c(k1, λ1)
χ0j
˜−R (p) τ (k2, λ2)
τ+1 (k3)
`c (k1, λ1)
χ0j
˜−R (p) τ c (k2, λ2)
τ+1 (k3)
`c (k1, λ1)
χ0j
Figure 48: Feynman diagrams for the three-body slepton decays ˜−R → `−τ+τ−1 (top row) and˜−R → `−τ−τ+
1 (bottom row) in the MSSM.
6.16 Three-body slepton decays ˜−R → `−τ± τ∓1 for ` = e, µ
In this subsection, we consider the three-body decays of sleptons through a virtual neutralino.
The usual assumption in supersymmetric phenomenology is that these decays will have a very
small branching fraction, because a two-body decay to a lighter neutralino and lepton is al-
ways open. However, in Gauge Mediated Supersymmetry Breaking models with a non-minimal
messenger sector, the sleptons can be lighter than the lightest neutralino [93, 94]. In that case,
the mostly right-handed smuon and selectron, µR and eR, will decay by ˜−R → `−τ±τ∓1 . The
lightest stau mass eigenstate, τ±1 , is a mixture of the weak eigenstates τ±L and τ±R , as described
in Appendix I.4:
τ−1 = c∗τ τ−R + s∗τ τ
−L , (6.197)
and τ+1 = (τ−1 )∗, while the µR and eR are taken to be unmixed.
First consider the decay ˜−R → `−τ+τ−1 , which proceed by the diagrams in the first row
of Fig. 48. The momenta and polarizations of the particles are also indicated on the diagram.
Using the Feynman rules of Fig. 78, we find that the amplitudes of these two diagrams, for each
neutralino Nj exchanged, are:
iM1 = (−ia ˜j)(−iaτ∗j ) y1
[ −i(p− k1)·σ(p− k1)2 −m2
Nj
]x2, (6.198)
iM2 = (−ia ˜j)(−ibτj ) y1
[ imNj
(p− k1)2 −m2Nj
]y2. (6.199)
84
where
a˜j =√
2g′N∗j1, (6.200)
aτj = YτN∗j3sτ +
√2g′N∗
j1cτ , (6.201)
bτj = YτN∗j3c
∗τ −
1√2(gN∗
j2 + g′N∗j1)s
∗τ . (6.202)
The spinor wavefunction factors are y1 = y(~k1, λ1), y2 = y(~k2, λ2), and x2 = x(~k2, λ2).
In the following, we will use the kinematic variables
z` ≡ 2p·k1/m2˜R
= 2E`/m˜R, zτ ≡ 2p·k2/m
2˜R
= 2Eτ/m˜R, (6.203)
rNj ≡ mNj/m˜
R, rτ ≡ mτ1/m˜
R, (6.204)
rτ ≡ mτ/m˜R, r` ≡ m`/m˜
R. (6.205)
The total amplitude then can be written as
M =
4∑
j=1
[cjy1(p− k1)·σx2 + djy1y2] (6.206)
where
cj = −a ˜jaτ∗j /[m
2`R
(r2Nj− 1 + z`)], (6.207)
dj = a˜jbτjmNj
/[m2`R
(r2Nj− 1 + z`)]. (6.208)
We consistently neglect the electron and muon Yukawa couplings (so r` = 0) in the matrix
elements, but not below in the kinematic integration over phase space, where the muon mass
can be important.
Now using eqs. (2.30) and (2.31), we find
|M|2 =∑
j,k
[cjc
∗k y1(p− k1)·σx2 x2(p− k1)·σy1 + djd
∗ky1y2 y2y1
+cjd∗ky1(p− k1)·σx2 y2y1 + c∗jdkx2(p− k1)·σy1 y1y2
]. (6.209)
Now summing over the lepton spins using eqs. (3.56)-(3.59),
∑
λ1,λ2
|M|2 =∑
j,k
[cjc
∗kTr[(p− k1)·σk2 ·σ(p− k1)·σk1 ·σ] + djd
∗kTr[k2 ·σk1 ·σ]
−cjd∗kmτTr[(p− k1)·σk1 ·σ]− c∗jdkmτTr[(p− k1)·σk1 ·σ]. (6.210)
85
Taking the traces using eqs. (2.43) and (2.44) yields
∑
spins
|M|2 =∑
j,k
cjc
∗k[4k1 ·(p− k1)k2 ·(p− k1)− 2k1 ·k2(p− k1)
2] + 2djd∗kk1 ·k2
−4Re[cjd∗k]mτk1 ·(p− k1)
(6.211)
=∑
j,k
cjc
∗km
4˜R[(1− z`)(1 − zτ )− r2
τ + r2τ ]
+djd∗km
2˜R(z` + zτ − 1 + r2
τ − r2τ )− 2Re[cjd∗k]mτm
2˜Rz`
(6.212)
The differential decay rate for ˜−R → `−τ+τ−1 then follows:
d2Γ
dz`dzτ=
m˜R
256π3
(∑
spins
|M|2)
(6.213)
The total decay rate in that channel can be found by integrating over z`, zτ , with the limits (see
for example ref. [75]):
2r` < z` < 1 + r2` − (rτ + rτ )
2, (6.214)
zτ <>
1
2(1− z` + r2` )
[(2− z`)(1 + r2
` + r2τ − r2τ − z`)± (z2` − 4r2
` )1/2λ1/2(1 + r2
` − z`, r2τ , r2τ )].
(6.215)
Now we turn to the competing decay ˜−R → `−τ−τ+1 , with diagrams appearing in the second
row of Fig. 48. By appealing again to the Feynman rules of Fig. 77, we find that the amplitude
has exactly the same form as in eqs. (6.198) and (6.199), except now with a τj ↔ bτj . Therefore,
the entire previous calculation goes through precisely as before, but now with
cj = −a ˜jbτ∗j /[m
2`R
(r2Nj− 1 + z`)], (6.216)
dj = a˜jaτjmNj
/[m2`R
(r2Nj− 1 + z`)]. (6.217)
The differential decay widths found above can be integrated to find the total decay widths. The
results agree with ref. [95], except that the signs of the coefficient c(3)ij and c
(4)ij in that paper are
incorrect and should be flipped. (Also, the notations for the sfermion mixing angle are different
in that paper.) If m ˜R−mτ1 −mτ is not too large, the resulting decays can have a macroscopic
length in a detector, and the ratio of the two decay modes can provide an interesting probe of
the supersymmetric Lagrangian.
6.17 Neutralino decay to photon and Goldstino: Ni → γG
The Goldstino G is a Weyl fermion that couples to the neutralino and photon fields according
to the non-renormalizable Lagrangian term [96]:
L = −ai2
(χ0i σµσρσν∂
µG) (∂νAρ − ∂ρAν) + c.c. (6.218)
86
χ0i (p, λN )
G (kG, λG)
γ (kγ , λγ)
χ0i (p, λN )
G (kG, λG)
γ (kγ , λγ)
Figure 49: The two Feynman diagrams for Ni → γG in supersymmetric models with a lightGoldstino.
Here χ0i is the left-handed two-component fermion field that corresponds to the neutralino Ni
particle, G is the two-component fermion field corresponding to the (nearly) massless Goldstino,
and the effective coupling is
ai ≡1√
2〈F 〉(N∗
i1 cos θW +N∗i2 sin θW ), (6.219)
whereNij the mixing matrix for the neutralinos [see eq. (I.23)], and 〈F 〉 is the F -term expectation
value associated with supersymmetry breaking. Therefore Ni can decay to γ plus G through
the diagrams shown in Fig. 49, with amplitudes:
iM1 = iai2xNkG ·σ (ε∗ ·σ kγ ·σ − kγ ·σ ε∗ ·σ) xG , (6.220)
iM2 = −ia∗i
2yNkG ·σ (ε∗ ·σ kγ ·σ − kγ ·σ ε∗ ·σ) yG . (6.221)
Here xN ≡ x(~p, λN ), yN ≡ y(~p, λN ), and xG ≡ x(~kG, λG), yG ≡ y(~kG, λG), and ε∗ = ε∗(~kγ , λγ)
are the external wavefunction factors for the neutralino, Goldstino, and photon, respectively. Us-
ing the on-shell condition kγ ·ε∗ = 0, we have kγ ·σε∗ ·σ = −ε∗ ·σkγ ·σ and kγ ·σε∗ ·σ = −ε∗ ·σkγ ·σfrom eqs. (2.39) and (2.40). So we can rewrite the total amplitude as
M =M1 +M2 = xNAxG + yNByG (6.222)
where
A = ai kG ·σ ε∗ ·σ kγ ·σ, (6.223)
B = −a∗i kG ·σ ε∗ ·σ kγ ·σ. (6.224)
The complex square of the matrix element is therefore
|M|2 = xNAxGxGAxN + yNByGyGByN + xNAxGyGByN + yNByGxGAxN , (6.225)
where A and B are obtained from A and B by reversing the order of the σ and σ matrices
and taking the complex conjugates of ai and ε. [See the discussion surrounding eqs. (4.23) and
(4.24).]
87
Summing over the Goldstino spins using eqs. (3.56)-(3.59) now yields:
∑
λG
|M|2 = xNAkG ·σAxN + yNBkG ·σByN . (6.226)
(The A, B and A, B cross terms vanish because of mG = 0.) Now averaging over the neutralino
spins using eqs. (3.56) and (3.57), we find
1
2
∑
λN ,λG
|M|2 =1
2Tr[AkG ·σAp·σ] +
1
2Tr[BkG ·σBp·σ]. (6.227)
=1
2|ai|2Tr[ε∗ ·σ kγ ·σ kG ·σ kγ ·σ ε·σ kG ·σ p·σ kG ·σ] + (σ ↔ σ). (6.228)
Now use
kγ ·σ kG ·σ kγ ·σ = 2kG ·kγ kγ ·σ, (6.229)
kG ·σ p·σ kG ·σ = 2kG ·p kG ·σ, (6.230)
which follow from eq. (2.41), and the corresponding identities with σ ↔ σ, to obtain:
1
2
∑
λN ,λG
|M|2 = 2|ai|2(kG ·kγ)(kG ·p)Tr[ε∗ ·σ kγ ·σ ε·σ kG ·σ] + (σ ↔ σ). (6.231)
Applying the photon spin sum identity
∑
λγ
εµεν∗ = −gµν (6.232)
and the trace identities eq. (2.44) and (2.45), we get
1
2
∑
λγ ,λN ,λG
|M|2 = 16|ai|2(kG ·kγ)2(kG ·p) = 2|ai|2m6Ni. (6.233)
So, the decay rate is [97]:
Γ(Ni → γG) =1
16πm eNi
1
2
∑
λγ ,λN ,λG
|M|2 = |Ni1 cos θW +Ni2 sin θW |2
m5eNi
16π|〈F 〉|2 . (6.234)
6.18 Gluino pair production from gluon fusion: gg → gg
In this subsection we will compute the cross-section for the process gg → gg. The relevant
Feynman diagrams are shown in Figure 50. The initial state gluons have SU(3)c adjoint rep-
resentation indices a and b, with momenta p1 and p2 and polarization vectors εµ1 = εµ(~p1, λ1)
and εµ2 = εµ(~p2, λ2) respectively. The final state gluinos carry adjoint representation indices c
and d, with momenta k1 and k2 and wavefunction spinors x1 = x(~k1, λ′1) or y1 = y(~k1, λ
′1) and
x2 = x(~k2, λ′2) or y2 = y(~k2, λ
′2), respectively.
88
ga (p1, λ1)
gb (p2, λ2)
gc (k1, λ′1)
gd (k2, λ′2)
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
ga
gb
gc
gd
ge
Figure 50: The ten Feynman diagrams for gg → gg. The momentum and spin polarizationassignments are indicated on the first diagram.
89
The Feynman rules for the gluino couplings in SUSYQCD are given in Fig. 79. For the two
s-channel amplitudes, we obtain:
iMs =(−gsfabe[gµν(p1 − p2)ρ + gνρ(p1 + 2p2)µ − gµρ(2p1 + p2)ν ]
)(−igρκs
)
εµ1εν2
[(−gsf cde) x1σκy2 + (gsf
dce) y1σκx2
]. (6.235)
The first factor is the Feynman rule for the three-gluon interaction of standard QCD, and
the second factor is the gluon propagator. The next four (t-channel) diagrams have a total
amplitude:
iMt =(−gsf ceaεµ1
)(−gsf edbεν2
)x1σµ
[i(k1 − p1)·σ
(k1 − p1)2 −m2g
]σνy2
+(gsf
ecaεµ1)(gsf
debεν2)y1σµ
[i(k1 − p1)·σ
(k1 − p1)2 −m2g
]σν x2
+(−gsf ceaεµ1
)(gsf
debεν2)x1σµ
[img
(k1 − p1)2 −m2g
]σν x2
+(gsf
ecaεµ1)(−gsf edbεν2
)y1σµ
[img
(k1 − p1)2 −m2g
]σνy2. (6.236)
Finally, the u-channel Feynman diagrams result in:
iMu =(−gsf edaεµ1
)(−gsf cebεν2
)x1σν
[i(k1 − p2)·σ
(k1 − p2)2 −m2g
]σµy2
+(gsf
deaεµ1)(gsf
ecbεν2)y1σν
[i(k1 − p2)·σ
(k1 − p2)2 −m2g
]σµx2
+(gsf
deaεµ1)(−gsf cebεν2
)x1σν
[img
(k1 − p2)2 −m2g
]σµx2
+(−gsf edaεµ1
)(gsf
ecbεν2)y1σν
[img
(k1 − p2)2 −m2g
]σµy2. (6.237)
We choose to work with real polarization vectors ε1, ε2. Since they must both be orthogonal to
the initial-state collision axis in the center of momentum frame, we have:
ε1 ·ε1 = ε2 ·ε2 = −1 (6.238)
ε1 ·p1 = ε2 ·p1 = ε1 ·p2 = ε2 ·p2 = 0, (6.239)
ε1 ·k2 = −ε1 ·k1, ε2 ·k2 = −ε2 ·k1, (6.240)
and the sums over gluon polarizations will be accomplished by:
∑
λ1
εµ1εν1 =
∑
λ2
εµ2εν2 = gµν + 2 (pµ1p
ν2 + pµ2p
ν1) /s. (6.241)
90
Before taking the complex square of the amplitude, it is convenient to rewrite the last two
terms in each of eqs. (6.236) and (6.237) by using the identities [see eq. (3.12)]:
mgx1 = y1σ ·k1, mgy1 = x1σ ·k1. (6.242)
Using eqs. (2.41) and (2.42), the resulting total matrix element is then reduced to a sum of
terms that each contain exactly one σ or σ matrix. We define convenient factors:
Gs ≡ g2sf
abef cde/s, (6.243)
Gt ≡ g2sf
acef bde/(t−m2g), (6.244)
Gu ≡ g2sf
adef bce/(u−m2g). (6.245)
where the usual Mandelstam variables are:
s = (p1 + p2)2 = (k1 + k2)
2, (6.246)
t = (k1 − p1)2 = (k2 − p2)
2, (6.247)
u = (k1 − p2)2 = (k2 − p1)
2. (6.248)
Then the total amplitude is (noting that the gluon polarizations were chosen real):
M =Ms +Mt +Mu = x1a·σy2 + y1a∗ ·σx2, (6.249)
where
aµ ≡ −(Gt +Gs)ε1 ·ε2 pµ1 − (Gu −Gs)ε1 ·ε2 pµ2 − 2Gtk1 ·ε1 εµ2 − 2Guk1 ·ε2 εµ1−iεµνρκε1νε2ρ(Gtp1 −Gup2)κ. (6.250)
Squaring the amplitude using eqs. (2.31) and (2.32), we get:
|M|2 = x1a·σy2y2a∗ ·σx1 + y1a
∗ ·σx2x2a·σy1 + x1a·σy2x2a·σy1 + y1a∗ ·σx2y2a
∗ ·σx1. (6.251)
Now summing over the gluino spins using eqs. (3.56)-(3.59), we find:
∑
λ′1,λ′2
|M|2 = Tr[a·σk2 ·σa∗ ·σk1 ·σ] + Tr[a∗ ·σk2 ·σa·σk1 ·σ]
−m2gTr[a·σa·σ]−m2
gTr[a∗ ·σa∗ ·σ]. (6.252)
Taking the traces with eqs. (2.43)-(2.45) yields:
∑
λ′1,λ′2
|M|2 = 8Re[a·k1a∗ ·k2]− 4a·a∗ k1 ·k2 − 4iεµνρκk1µk2νaρa
∗κ − 4m2
gRe[a2]. (6.253)
Now plugging in eqs. (6.250), we obtain:
∑
λ′1,λ′2
|M|2 = 2(t−m2g)(u−m2
g)[(Gt +Gu)2 + 4(Gs +Gt)(Gs −Gu)(ε1 ·ε2)2]
+16(Gt +Gu)[Gs(t− u) +Gt(t−m2g) +Gu(u−m2
g)](ε1 ·ε2)(k1 ·ε1)(k1 ·ε2)
−32(Gt +Gu)2(k1 ·ε1)2(k1 ·ε2)2. (6.254)
91
The sums over gluon polarizations can be done using eq. (6.241), which implies:
∑
λ1,λ2
1 = 4,∑
λ1,λ2
(ε1 ·ε2)2 = 2, (6.255)
∑
λ1,λ2
(ε1 ·ε2)(k1 ·ε1)(k1 ·ε2) = m2g − (t−m2
g)(u−m2g)/s (6.256)
∑
λ1,λ2
(k1 ·ε1)2(k1 ·ε2)2 =(m2g − (t−m2
g)(u−m2g)/s
)2. (6.257)
Also, we can sum over colors using f abef cdefabe′f cde
′= 2fabef cdeface
′f bde
′= N2
c (N2c − 1) = 72,
so:
∑
colors
G2s =
72g4s
s2,
∑
colors
G2t =
72g4s
(t−m2g)
2, (6.258)
∑
colors
G2u =
72g4s
(u−m2g)
2,
∑
colors
GsGt =36g4
s
s(t−m2g), (6.259)
∑
colors
GsGu = − 36g4s
s(u−m2g),
∑
colors
GtGu =36g4
s
(t−m2g)(u−m2
g). (6.260)
Putting the factors together, and averaging over the initial state colors and spins, we have:
dσ
dt=
1
16πs2
(1
64
∑
colors
1
4
∑
spins
|M|2)
(6.261)
=9πα2
s
4s4
[2(t−m2
g)(u−m2g)− 3s2 − 4m2
gs+s2(s+ 2m2
g)2
(t−m2g)(u−m2
g)
−4m4
gs4
(t−m2g)
2(u−m2g)
2
], (6.262)
which agrees with the result of [87, 98] (after some rearrangment). Note that in the center-of-
momentum frame, the Mandelstam variable t is related to the scattering angle θ between an
initial-state gluon and a final-state gluino by:
t = m2g +
(cos θ
√1− 4m2
g/s− 1)s/2. (6.263)
Since the final state has identical particles, the total cross-section can now be obtained by:
σ =1
2
∫ t+
t−
dσ
dtdt (6.264)
where t± are obtained by inserting cos θ = ±1 into eq. (6.263)..
6.19 R-parity violating stau decay: τ+R → e+νµ
Next we consider the decay of a right-handed scalar tau via an R-parity violating LLE coupling.
This is particularly relevant for a scalar tau LSP [99,100] and resonant slepton production [101].
92
τ+R
e(ke, λe)
νµ(kνµ , λνµ)
Figure 51: Feynman diagram for the R-parity violating decay τ+R → e+νµ
The Feynman diagram is shown in Fig. 6.19, where we have also defined the momenta and the
helicities of the fermions. The amplitude is given by
iM = −iλyeyνµ , (6.265)
where we have denoted the external wave functions as ye ≡ y(~ke, λe), and yνµ ≡ y(~kνµ , λνµ),
respectively. We have also written the R-parity-violating Yukawa coupling as λ ≡ λ123 (see
Appendix J). Using eq. (2.30), the amplitude squared is
|M|2 = |λ|2yeyνµ yνµ ye. (6.266)
Summing over the fermion spins using eqs. (3.57) gives:
∑
λe,λνµ
|M|2 = |λ|2Tr[ke ·σ kνµ ·σ] = |λ|2m2τR, (6.267)
where in the last step we have used the trace formula eq. (2.43), and neglected the mass of the
electron and the neutrino. The total decay rate is then given by
Γ =1
16πmτR
( ∑
λe,λνµ
|M|2)
=|λ|216π
mτR , (6.268)
which agrees with the computation in [102–104]. Completely analogously we can obtain the
total rate for the decays νµ → τ−e+ and e−L → τ−νµ, which proceed via the same operator, by
replacing mτR → (meL ,mνµ), respectively.
In general the two-body decay rate of a sfermion f via the LQD or UDD interaction is
given by
Γ(f → f1f2) =C|λ|216π
mf , (6.269)
where we have neglected the masses m1,2 of the final state fermions. The factor C denotes the
color factor. For the slepton decays via LQD which are summed over the final-state quark colors,
C = Nc = 3. For the squark decays via LQD where the initial state color is averaged over and
the final-state color is summed, C = 1. For the squark decays via UDD, C = (Nc − 1)! = 2. In
realistic cases, one must also include the effects of mixing for the third-family sfermions, which
we have omitted here for simplicity.
93
χ0i (pi, λi) u (ku, λu)
dc (kd, λd)
µ (kµ, λµ)
µL
χ0i (pi, λi) µ (kµ, λµ)
u (ku, λu)
dc (kd, λd)
dR
χ0i (pi, λi) µ (kµ, λµ)
dc (kd, λd)
u (ku, λu)
uL
Figure 52: Feynman diagrams for the R-parity violating decay Ni → µ−ud.
6.20 R-parity violating neutralino decay: Ni → µ−ud
Next we consider the R-parity violating three-body decay of a neutralino Ni → µ−ud, which
arises via the superpotential term λ′211LµQ1D1. This is of particular interest when the neutralino
is the LSP, since it determines the final-state signatures [105,106]. The three Feynman diagrams
are shown in Fig. 52, including the definitions of the momenta and helicities. We have neglected
sfermion mixing, i.e. we assume µL, uL, and dR are mass eigenstates. Using the Feynman rules
given in Figs. 76 (or 78) and 83, we obtain the amplitudes
iM1 =(iλ′∗) [ i√
2(gNi2 + g′Ni1)
] [i
(pi − kµ)2 −m2µL
]yixµxuxd (6.270)
iM2 =(iλ′∗)[− i√
2
3g′Ni1
][i
(pi − kd)2 −m2dR
]yixdxµxu (6.271)
iM3 =(iλ′∗) [− i√
2(gNi2 + g′Ni1/3)
] [i
(pi − ku)2 −m2uL
]yixuxdxµ (6.272)
Here we have denoted the external wave functions as yi ≡ y(~pi, λi), xµ ≡ x(~kµ, λµ), xu ≡x(~ku, λu), xd ≡ x(~kd, λd), and written λ′ ≡ λ′211 . In the following, we will neglect all of the
94
final-state fermion masses. The results will be expressed in terms of the kinematic variables
zµ ≡ 2pi ·kµ/m2Ni
= 2Eµ/mNi, (6.273)
zd ≡ 2pi ·kd/m2Ni
= 2Ed/mNi, (6.274)
zu ≡ 2pi ·ku/m2Ni
= 2Eu/mNi(6.275)
which satisfy zµ + zd + zu = 2. Then we can rewrite the total matrix element as:
M = c1yixµxuxd + c2yixdxµxu + c3yixuxdxµ (6.276)
where
c1 ≡1√2λ′∗(gNi2 + g′Ni1)/[m
2µL −m
2Ni
(1− zµ)], (6.277)
c2 ≡ −√
2
3λ′∗g′Ni1/[m
2dR−m2
Ni(1− zd)], (6.278)
c3 ≡ −1√2λ′∗(gNi2 + g′Ni1/3)/[m
2uL−m2
Ni(1− zu)]. (6.279)
Before squaring the amplitude, it is convenient to use the Fierz identity (2.54) to reduce the
number of terms:
M = (c1 − c3)yixµxuxd + (c2 − c3)yixdxµxu. (6.280)
Now, using eq. (2.30), we obtain
|M|2 = |c1 − c3|2yixµxµyixuxdxdxu + |c2 − c3|2yixdxdyixµxuxuxµ−2Re[(c1 − c3)(c∗2 − c∗3)yixµxµxuxuxdxdyi] , (6.281)
where eq. (2.47) was used on the last term. Now summing over the fermion spins using
eqs. (3.56)-(3.59), we obtain:
∑
spins
|M|2 = |c1 − c3|2Tr[kµ ·σpi ·σ]Tr[kd ·σku ·σ] + |c2 − c3|2Tr[kd ·σpi ·σ]Tr[ku ·σkµ ·σ]
−2Re[(c1 − c3)(c∗2 − c∗3)Tr[kµ ·σku ·σkd ·σpi ·σ]
]. (6.282)
Applying the trace formulas (2.43) and (2.45), we obtain
∑
spins
|M|2 = 4|c1 − c3|2pi ·kµ kd ·ku + 4|c2 − c3|2pi ·kd kµ ·ku
−4Re[(c1 − c3)(c∗2 − c∗3)](kµ ·ku pi ·kd + pi ·kµ kd ·ku − kµ ·kd pi ·ku) (6.283)
= m4Ni
[|c1|2zµ(1− zµ) + |c2|2zd(1− zd) + |c3|2zu(1− zu)
−2Re[c1c∗2](1− zµ)(1− zd)− 2Re[c1c
∗3](1− zµ)(1− zu)
−2Re[c2c∗3](1− zd)(1− zu)
](6.284)
95
where in the last equality we have used eqs. (6.273)-(6.275) and
2kµ ·kd = (1− zu)m2Ni, 2kµ ·ku = (1− zd)m2
Ni, 2kd ·ku = (1− zµ)m2
Ni. (6.285)
The differential decay rate follows:
d2Γ
dzµdzd=NcmNi
256π3
(1
2
∑
spins
|M|2), (6.286)
where a factor of Nc = 3 has been included for the sum over colors, a factor of 1/2 to average
over the neutralino spin, and the kinematic limits are
0 < zµ < 1, (6.287)
1− zµ < zd < 1. (6.288)
In the limit of heavy sfermions, the integrations over zd and then zµ are simple, with the result
for the total decay width:
Γ =Ncm
5Ni
6144π3
(|c′1|2 + |c′2|2 + |c′3|2 −Re[c′1c
′∗2 + c′1c
′∗3 + c′2c
′∗3 ]), (6.289)
where the c′i are obtained from ci of eqs. (6.277)-(6.279) by neglecting m2Ni
in the denominators.
Our results agree with the complete computation given in [103,104,107], where also the complete
mixing was included. Earlier calculations with some simplifications are given in [106, 108].
6.21 Top-quark condensation from a Nambu-Jona-Lasinio model gap equa-tion
The previous examples have involved renormalizable field theories. However, there are cases in
which it is preferable to use effective four-fermion interactions. The obvious historical example
is the use of the four-fermion Fermi theory of weak decays. This has been superseded by a
more complete and accurate theory of the weak interactions, but is still useful for leading-order
calculations of low-energy processes. Another case of some interest is the use of strong coupling
four-fermion interactions to drive symmetry breaking via a Nambu-Jona-Lasinio model [109], as
in the top-quark condensate approach [110]- [113] to electroweak symmetry breaking.
Consider an effective four-fermion Lagrangian involving the top quark [111], written in
two-component fermion form as:
L = itσµ∂µt+ itcσµ∂µtc +
G
Λ2(ttc)(ttc). (6.290)
Here the Standard Model gauge interactions have been suppressed; the quantities within paren-
theses are color singlets. Note also that there is no top quark Yukawa coupling to a Higgs scalar
boson, nor a top quark mass term, which would normally appear in the form −mt(ttc + ttc).
96
t, i, α
tc, j, β
t, k, α
tc, n, β
iG
Λ2δji δ
knδβαδ
βα
Figure 53: Feynman rule for the four-fermion interaction in the top-quark condensate model.The indices i, j, k, n = 1, 2, 3 are for color in the fundamental representation of SU(3), and theindices α, β, α, β are two-component spinor indices.
=
Figure 54: The Nambu-Jona-Lasinio gap equation for a possible dynamically generated top-quark mass mt.
Instead, the effective top quark mass is supposed to be driven by a non-perturbatively large and
positive dimensionless coupling G, with Λ the cutoff scale at which G arises from some more
fundamental physics such as topcolor [113].
The Feynman rule for the four-fermion interaction can be derived from the mode expansion
results of section 3, and is given in Fig. 53. The resulting gap equation for the dynamically
generated top quark mass is shown in Fig. 54. Evaluating this using the Feynman rules of figs. 4
and 5, one finds:
−imtδji δβα = (−1)
∫ Λ d4k
(2π)4
(iG
Λ2δji δ
knδβαδ
βα
) (δnk δ
αβ
imt
k2 −m2t + iε
). (6.291)
Here i, j, k, n are color indices of the fundamental representation of SU(3), and α, β, α, β are
two-component spinor indices. The factor of (−1) on the right-hand side is due to the presence
of a fermion loop.
Euclideanizing the loop integration over kµ by k2 → −k2E and
∫d4k → i
∫d4kE , and then
rewriting the integration in terms of x = k2E , this amounts to [111]:
mt =2NcGmt
16π2Λ2
∫ Λ2
0dx/(1 +m2
t /x) (6.292)
=3Gmt
8π2[1− (m2
t /Λ2) ln(Λ2/m2
t ) + . . .], (6.293)
where Nc = 3 is the number of colors, and a factor of two arose from the sum over dotted spinor
indices of δβαδαβ.
For small or negative G, only the trivial solution mt = 0 is possible. However, for G ≥Gcritical = 8π2/3 ≈ 26, there is always a positive solution for m2
t /Λ2 [111]. It is now known
97
that this minimal version of the model cannot explain the top quark mass and the observed
features of electroweak symmetry breaking, but extensions of it may be viable [114].
6.22 Electroweak vector boson self-energies from fermion loops
In this subsection, we consider the contributions to the self-energy functions of the Standard
Model electroweak vector bosons coming from quark and lepton loops. (For a derivation of
equivalent results in the four-component fermion formalism, see for example section 21.3 of [76].)
The independent self-energies are given by ΠWWµν , ΠZZ
µν , ΠγZµν = ΠZγ
µν , and Πγγµν , as shown in figs. 55
and 56. In each case, iΠµν is equal to the sum of Feynman diagrams for two-point functions
with amputated external legs, and is implicitly a function of the external momentum pµ.
First consider the self-energy function for the W boson, shown in Fig. 55. The W boson
only couples to left-handed fermions, so there is only one Feynman diagram for each Standard
model weak isodoublet. Taking the external momentum flowing from left to right to be p, and
the loop momentum flowing counterclockwise in the upper fermion line (f) to be k, we have
from the Feynman rules of Fig. 67:
iΠWWµν = (−1)µ2ε
∫ddk
(2π)d
∑
(f,f ′)
Nfc Tr
[(−i g√
2σµ
)( ik ·σk2 −m2
f
)(−i g√
2σν
)( i(k + p)·σ(k + p)2 −m2
f ′
)].
(6.294)
Here µ is a regularization scale for dimensional regularization in d ≡ 4 − 2ε dimensions. The
sum in eq. (6.294) is over the six isodoublet pairs (f, f ′) = (e, νe), (µ, νµ), (τ, ντ ), (d, u), (s, c),
iΠWWµν (p) = W+ W+
f
f ′
p p
µ ν
Figure 55: Contributions to the self-energy function for the W boson in the Standard Model,from loops involving the left-handed quark and lepton pairs (f, f ′) = (e, νe), (µ, νµ), (τ, ντ ),(d, u), (s, c), and (b, t). The momentum of the positively charged W + flows from left to right.
iΠV V ′
µν =
f
f
µ ν
V V ′
+
f c
f cµ ν
V V ′
+
f
f
f c
f cµ ν
V V ′
+
f c
f c
f
fµ ν
V V ′
Figure 56: Contributions to the diagonal and off-diagonal self-energy functions for the neutralvector bosons V, V ′ = γ, Z in the Standard Model, from loops involving the three generationsof leptons and quarks: f = e, νe, µ, νµ, τ, ντ , d, u, s, c, b, t.
98
and (b, t) with CKM mixing neglected, and
Nfc =
3 , f = quarks ,
1 , f = leptons .(6.295)
The first factor of (−1) in eq. (6.294) is due to the presence of a closed fermion loop. The trace
is taken over the two-component dotted spinor indices. Using eq. (A.25), it follows that
ΠWWµν =
g2
32π2
∑
f
Nfc Iµν(m
2f ,m
2f ′) , (6.296)
where we have defined
Iµν(x, y) = i(16π2)µ2ε
∫ddk
(2π)d4kµkν + 2kµpν + 2kνpµ − 2k ·(k + p) gµν
(k2 − x)[(k + p)2 − y] . (6.297)
This integral can be evaluated by the standard dimensional regularization methods [76, 115],
with the result:
Iµν(x, y) = (p2gµν − pµpν)I1(p2;x, y) + gµνI2(p2;x, y), (6.298)
where, after neglecting terms that vanish as ε→ 0,
I1(s;x, y) = − 2
3ε+
2
3s2
(2x− 2y − s)A(x) + (2y − 2x− s)A(y)
+[2(x− y)2 − s(x+ y)− s2
]B(s;x, y)− s(x+ y) + s2/3
, (6.299)
I2(s;x, y) =x+ y
ε− 1
s
(x− y)
[A(x)−A(y)
]+[(x− y)2 − s(x+ y)
]B(s;x, y)
. (6.300)
The functions
A(x) ≡ x ln(x/Q2)− x, (6.301)
B(s;x, y) ≡ −∫ 1
0dt ln
(tx+ (1− t)y − t(1− t)s− iε
Q2
), (6.302)
are the finite parts of one-loop Passarino-Veltman functions, with the renormalization scale Q
related to the regularization scale µ by the modified minimal subtraction relation
µ2 = Q2eγ/4π, (6.303)
where γ = 0.577216 . . . is Euler’s constant.
The photon and Z boson have mixed self-energy functions, defined in Fig. 56. Applying
99
the pertinent Feynman rules from Fig. 67, we obtain:
iΠV V ′
µν = (−1)µ2ε
∫ddk
(2π)d
∑
f
Nfc Tr
(−iGfV σµ
)( ik ·σk2 −m2
f
)(−iGfV ′σν
)( i(k + p)·σ(k + p)2 −m2
f
)
+(−iGfcV σµ
)( ik ·σk2 −m2
f
)(−iGfcV ′σν
)( i(k + p)·σ(k + p)2 −m2
f
)
+(−iGfV σµ
)( imf
k2 −m2f
)(iGf
c
V ′σν
)( imf
(k + p)2 −m2f
)
+(−iGfcV σµ
)( imf
k2 −m2f
)(iGfV ′σν
)( imf
(k + p)2 −m2f
), (6.304)
where V and V ′ can each be either γ or Z, and∑
f is taken over the 12 Standard Model fermions.
The corresponding V ff and V f cf c couplings are:34
Gfγ = −Gfcγ = eQf , (6.305)
GfZ =g
cW(T f3 − s2WQf ), Gf
c
Z =g
cWs2WQf . (6.306)
The four terms in eq. (6.304) correspond to the four diagrams in Fig. 56, in the same order.
The first two terms in eq. (6.304) are computed exactly as for ΠWWµν , while in the last two
terms we use eq. (A.3) to compute the trace. It follows that the neutral electroweak vector
boson self-energy function matrix, after dropping terms that vanish as ε→ 0, is given by
ΠV V ′
µν =1
16π2
∑
f
Nfc
[(GfVG
fV ′ +Gf
c
V Gfc
V ′)Iµν(m2f ,m
2f )
+gµν(GfVG
fc
V ′ +Gfc
V GfV ′)m
2fI3(m
2f ,m
2f )], (6.307)
where Iµν(x, y) was defined in eqs. (6.298)-(6.300), and we have defined the function
I3(x, y) = −i(16π2) µ2ε
∫ddk
(2π)d2
(k2 − x)[(k + p)2 − y] =2
ε+ 2B(p2;x, y). (6.308)
The photon self-energy function is a simple special case of eq. (6.307):
Πγγµν =
1
16π2
∑
f
2Nfc (eQf )
2[Iµν(m
2f ,m
2f )− gµνm2
fI3(m2f ,m
2f )]
(6.309)
=α
3π
∑
f
Nfc Q
2f
(p2gµν − pµpν
)− 1
ε+
1
3− 2
p2
[A(m2
f ) +m2f
]
−(
1 +2m2
f
p2
)B(p2;m2
f ,m2f )
, (6.310)
in agreement with the result given in, for example, eq. (7.90) of [76]. This formula satisfies
pµΠγγµν = pνΠγγ
µν = 0 as required by the Ward identity of QED, and is regular in the limit
p2 → 0.
34Note that there is no contribution from νce , νcµ, νcτ , which do not exist in the Standard Model.
100
In each of eqs. (6.296), (6.307), and (6.310), there are 1/ε poles, contained in the loop integral
functions. In the MS renormalization scheme, these poles are simply removed by counterterms,
which have no other effect.
In eqs. (6.294) and (6.304), we chose to write a σµ for the left vertex in the Feynman
diagram in each case. This is an arbitrary choice; we could also have chosen to use instead −σµfor the left vertex in any given diagram, as mentioned in the caption for Fig. 67. This would
have dictated the replacements σ ↔ −σ throughout the expression for the diagram, including
for the fermion propagators, as was indicated in Fig. 5. It is not hard to check that the result
after computing the spinor index traces is unaffected. Note that the contribution proportional
to εµνρκ from eq. (A.24) or eq. (A.25) vanishes; this is clear because the self-energy function is
symmetric under interchange of vector indices, and there is only one independent momentum in
the problem.
6.23 Self-energy and pole mass of the top quark
We next consider the one-loop calculation of the self-energy and the pole mass of the top quark in
the Standard Model, including the effects of the gauge interactions and the top and bottom quark
Yukawa couplings. As in Section 6.1, we treat this as a one-generation problem, neglecting CKM
mixing. Consequently, the corresponding Yukawa couplings Yt and Yb are real and positive (by a
suitable phase redefinition of the Higgs field35). Using the formalism of subsection 4.6 for Dirac
fermions, the independent 1PI self-energy functions are given by36 ΣLt, ΣRt and ΣDt (defined
in Fig. 22) as shown in Fig. 57. Note that in these diagrams, the physical top quark moves
from right to left, carrying momentum pµ. Then according to the general formula obtained in
eq. (4.71), the top-quark pole squared mass will be given by:
M2t − iΓtMt =
(mt + ΣDt)2
(1− ΣLT )(1− ΣRt), (6.311)
where mt is the tree-level mass. Working consistently to one-loop order, this yields
M2t − iΓtMt =
[m2t (1 + ΣLT + ΣRt) + 2mtΣDt
] ∣∣∣s=m2
t+iε. (6.312)
(It would be just as valid to substitute in s = M 2t + iε here, as two-loop order effects are
neglected.)
It remains to calculate the self-energy functions ΣLt, ΣRt and ΣDt. Two regularization pro-
cedures will be used simultaneously—the MS scheme based on dimensional regularization [116]
35As shown in Section 3.2, after the fermion-mass matrix diagonalization procedure, the tree-level fermionmasses are real and non-negative. If CKM mixing is neglected, it follows from eq. (H.13) that the correspondingdiagonal Yukawa couplings are real and positive if the phase of the Higgs field is chosen such that the neutralHiggs vacuum expectation value v > 0.
36Since the Yukawa couplings can be chosen real (in the one-generation model), ΣLt = ΣLt. Note that aftersuppressing the color degrees of freedom, ΣLt, ΣRt and ΣDt are one-dimensional matrices, so we do not employboldface letters in this case.
101
p
−ip·σΣLt
=
t t t
g, γ, Z
+
t b t
W+
+
t tc t
hSM, G0
+
t bc t
G+
p
−ip·σΣRt
=
tc tc tc
g, γ, Z
+
tc t tc
hSM, G0
+
tc b tc
G+
p
−iΣDt
=
tc tc t t
g, γ, Z
+
tc t tc t
hSM, G0
+
tc b bc t
G+
+
tc t
hSM
Figure 57: One-loop contributions to the 1PI self-energy functions for the top quark in theStandard Model. The external momentum of the physical top quark, pµ, flows from the right tothe left. The loop momentum kµ in the text is taken to flow clockwise. Spinor and color indicesare suppressed. The external legs are amputated. The last diagram contains one-loop tadpolecontributions.
and the DR scheme based on dimensional reduction [117]. This is accomplished by integrating
over the loop momentum in
d ≡ 4− 2ε (6.313)
dimensions, but with the vector bosons possessing
D ≡ 4− 2εδMS (6.314)
components, where
δMS ≡
1 for MS
0 for DR.(6.315)
In other words, the metric gµν appearing explicitly in the vector propagator is treated as four-
dimensional in DR, but as d-dimensional in MS. The renormalization scale Q is related to the
regularization scale µ in both cases by the modified minimal subtraction relation of eq. (6.303).
The calculation of the non-tadpole contributions to the self-energy functions will be per-
formed below in a general Rξ gauge, with a vector boson propagator as in fig. 6. There are
different ways to treat the tadpole contributions, corresponding to different choices for the Higgs
VEV around which the tree-level Lagrangian is expanded. If one chooses to expand around the
minimum of the tree-level Higgs potential, then there are no tree-level tadpoles, but there will
be non-zero contributions from the last diagram shown in fig. 57. (This corresponds to the
treatment given, for example, in ref. [118].) Alternatively, one can choose to expand around the
102
hSM + hSM = 0
Figure 58: The tree-level Higgs tadpole cancels against the one-loop Higgs tadpole, providedthat one expands around a Higgs VEV that minimizes the one-loop effective potential (ratherthan the tree-level Higgs potential, which would yield no tree-level tadpole).
Higgs VEV v that minimizes the one-loop Landau gauge37 effective potential. In that case, the
one-loop tadpole contribution is precisely cancelled by the tree-level Higgs tadpole, as shown
in fig. 58. Here, we have in mind the latter prescription; the calculation for the pole mass is
therefore complete without tadpole contributions provided that the tree-level top-quark mass is
taken to be
mt = Ytv, (6.316)
where Yt is the MS or DR Yukawa coupling, and v is the Higgs VEV at the minimum of the
one-loop effective potential in Landau gauge. To be consistent with this choice, ξ = 0 should
be taken in all formulas below that involve electroweak gauge bosons or Goldstone bosons.
(The gluon contribution is naturally independent of ξ because the gauge symmetry is unbroken,
providing a check of gauge-fixing invariance.) Nevertheless, for the sake of generality we will
keep the dependence on ξ in the computation of the individual non-tadpole self-energy diagrams
below.
Consider the one-loop calculation of the self-energy ΣLt, which is the sum of individual
diagram contributions ΣLt = [ΣLt]g+[ΣLt]γ+[ΣLt]Z+[ΣLt]W+[ΣLt]hSM+[ΣLt]G0+[ΣLt]G+ . First,
consider the diagrams involving exchanges of the scalars φ = hSM, G0, G±. These contributions
all have the same form
−ip·σ [ΣLt]φ = µ2ε
∫ddk
(2π)d(−iY ∗)
(i(k + p)·σ
(k + p)2 −m2f
)(−iY )
(i
k2 −m2φ
), (6.317)
where the loop momentum kµ flows clockwise, and the couplings and propagator masses are,
using the Feynman rules of figs. 68 and 69,
for φ = hSM : Y = Yt/√
2; mf = mt; m2φ = m2
hSM, (6.318)
for φ = G0 : Y = iYt/√
2; mf = mt; m2φ = ξm2
Z , (6.319)
for φ = G± : Y = Yb; mf = mb; m2φ = ξm2
W . (6.320)
37This procedure is considerably more involved outside of Landau gauge, because the propagators mix thelongitudinal components of the vector boson with the Nambu-Goldstone bosons for ξ 6= 0 if one expands arounda Higgs VEV that does not minimize the tree-level potential. This is the same reason the effective potential istraditionally calculated specifically in Landau gauge.
103
Multiplying both sides by p·σ and taking the trace over spinor indices using eq. (A.3), one finds
[ΣLt]φ = i|Y |2µ2ε
p2
∫ddk
(2π)dp·(k + p)
[(k + p)2 −m2f ][k
2 −m2φ]. (6.321)
Performing the loop momentum integration in the standard way [76, 115], and expanding in ε
up to constant terms, one finds that in each case
[ΣLt]φ = − 1
16π2|Y |2 IFS(s;m2
f ,m2φ). (6.322)
Here we have introduced some notation for the loop integral:
IFS(s;x, y) ≡ 1
2ε+ [(s+ x− y)B(s;x, y) +A(x) −A(y)]/2s , (6.323)
where the Passarino-Veltman functions A(x) and B(s;x, y) were defined in eqs. (6.301) and
(6.302). These functions depend on the renormalization scale Q, which is related to µ via
eq. (6.303). It can be checked that IFS(s;x, y) has a smooth limit as s→ 0.
Next, let us consider the contributions to ΣLt involving the vector bosons V = g, γ, Z,W .
These have the common form:
−ip·σ[ΣLt]V = µ2ε
∫ddk
(2π)d(−iGσµ)
(i(k + p)·σ
(k + p)2 −m2f
)(−iGσν)
( −ik2 −m2
V
)(gµν +
(ξ − 1)kµkν
k2 − ξm2V
), (6.324)
where again the loop momentum k flows clockwise, and, using the rules of figs. 67 and 79:
for V = g : G = gsTa; mf = mt, (6.325)
for V = γ : G = eQt; mf = mt, (6.326)
for V = Z : G = g(T t3 − s2WQt)/cW ; mf = mt, (6.327)
for V = W : G = g/√
2; mf = mb. (6.328)
In the case of gluon exchange (V = g), the T a are the SU(3)C generators (with color in-
dices suppressed). The adjoint representation index a is summed over, producing a factor of
the Casimir invariant (T aT a)ij = CF δij = 43δij . We now use σµ σρ σν g
µν = (2 − D)σρ [see
eq. (A.11)]; note that this introduces a difference between the MS and DR schemes. Also, we
use k ·σ(k + p)·σk ·σ = (k2 + 2k ·p)k ·σ − k2p·σ, which follows from eq. (2.42). One therefore
obtains, after multiplying by p·σ and taking the trace over spinor indices:
[ΣLt]V = −iG2µ2ε
p2
∫ddk
(2π)d1
[(k + p)2 −m2f ][k
2 −m2V ]
[(2−D)p·(k + p)
+(k2k ·p+ 2(k ·p)2 − k2p2
) (ξ − 1)
k2 − ξm2V
]. (6.329)
104
Performing the loop momentum integration, one finds in each case that
[ΣLt]V = − 1
16π2G2IFV (s;m2
f ,m2V ), (6.330)
where we have introduced the notation
IFV (s;x, y) =ξ
ε+ [(s+ x− y)B(s;x, y) +A(x)−A(y)]/s− δMS +
(s− x)[A(y)−A(ξy)]
+[(s− x)2 − y(s+ x)]B(s;x, y)− [(s− x)2 − ξy(s+ x)]B(s;x, ξy)/2ys, (6.331)
after dropping terms that vanish as ε→ 0. Combining the results of eqs. (6.322) and (6.330):
ΣLt = − 1
16π2
[(g2sCF + e2Q2
t
)IFV (m2
t ;m2t , 0) + [g(T t3 − s2WQt)/cW ]2IFV (m2
t ;m2t ,m
2Z)
+1
2g2IFV (m2
t ;m2b ,m
2W ) +
1
2Y 2t IFS(m2
t ;m2t ,m
2hSM
)
+1
2Y 2t IFS(m2
t ;m2t , ξm
2Z) + Y 2
b IFS(m2t ;m
2b , ξm
2W )], (6.332)
where we have now substituted s = m2t . It is useful to note that for massless gauge bosons,
IFV (x;x, 0) = ξ
[1
ε− ln(x/Q2) + 2
]+ 1− δMS. (6.333)
The contributions to ΣRt = [ΣRt]g + [ΣRt]γ + [ΣRt]Z + [ΣRt]hSM+ [ΣRt]G0 + [ΣRt]G± are
obtained similarly. [Note that there is no W boson contribution, since the right-handed top
quark is an SU(2)L singlet.] For the scalar exchange diagrams with φ = hSM, G0, G±, the
general form is:
−ip·σ[ΣRt]φ = µ2ε
∫ddk
(2π)d(−iY )
(i(k + p)·σ
(k + p)2 −m2f
)(−iY ∗)
(i
k2 −m2φ
). (6.334)
and so
[ΣRt]φ = − 1
16π2|Y |2 IFS(s;m2
f ,m2φ). (6.335)
Here the couplings and propagator masses for hSM and G0 are the same as in eqs. (6.318),
(6.319), but now, instead of eq. (6.320),
for φ = G± : Y = −Yt; mf = mb; m2φ = ξm2
W (6.336)
from fig. 69. For the contributions due to exchanges of vectors v = g, γ, Z, the general form is
−ip·σ[ΣRt]V = µ2ε
∫ddk
(2π)d(iGσµ)
(i(k + p)·σ
(k + p)2 −m2f
)(iGσν)
( −ik2 −m2
V
)(gµν +
(ξ − 1)kµkν
k2 − ξm2V
), (6.337)
105
where, using the rules of figs. 67 and 79:
for V = g : G = −gsT a; (6.338)
for V = γ : G = −eQt; (6.339)
for V = Z : G = gs2WQt/cW , (6.340)
with mf = mt in each case. Using σµ σρ σν gµν = (2 − D)σρ [see eq. (A.10)] and k ·σ(k +
p)·σk ·σ = (k2 + 2k ·p)k ·σ − k2p·σ [from eq. (2.41)], then multiplying by p·σ and taking the
trace over spinor indices [using eq. (A.3)], we obtain
[ΣRt]V = − 1
16π2G2IFV (s;m2
t ,m2V ) (6.341)
in terms of the same function appearing in eqs. (6.331) and (6.333). Adding up these contribu-
tions and taking s = m2t yields
ΣRt = − 1
16π2
[(g2sCF + e2Q2
t
)IFV (m2
t ;m2t , 0) + (g2Q2
t s4W/c
2W )IFV (m2
t ;m2t ,m
2Z)
+1
2Y 2t IFS(m2
t ;m2t ,m
2hSM
) +1
2Y 2t IFS(m2
t ;m2t , ξm
2Z) + Y 2
t IFS(m2t ;m
2b , ξm
2W )]. (6.342)
Next, consider the contributions to ΣDt = [ΣDt]g + [ΣDt]γ + [ΣDt]Z + [ΣDt]hSM+ [ΣDt]G0 +
[ΣDt]G± , ignoring the tadpole contribution for now. The diagrams involving the exchange of
scalars φ = hSM, G0, G± have the form:
−i[ΣDt]φ = µ2ε
∫ddk
(2π)d(−iY1)
(imf
(k + p)2 −m2f
)(−iY2)
(i
k2 −m2φ
), (6.343)
so that
[ΣDt]φ = imfY1Y2µ2ε
∫ddk
(2π)d1
[(k + p)2 −m2f ][k
2 −m2φ]
(6.344)
=1
16π2mfY1Y2IFS(s;m2
f ,m2φ) (6.345)
where we have introduced the notation:
IFS(s;x, y) ≡ −1
ε−B(s;x, y), (6.346)
after dropping terms that vanish as ε→ 0. The relevant couplings and masses are, from figs. 68
and 69:
for φ = hSM : Y1 = Y2 = Yt/√
2; mf = mt; m2φ = m2
hSM, (6.347)
for φ = G0 : Y1 = Y2 = iYt/√
2; ; mf = mt; m2φ = ξm2
Z , (6.348)
for φ = G± : Y1 = Yb; Y2 = −Yt; mf = mb; m2φ = ξm2
W . (6.349)
106
The contributions from vector boson exchanges are of the form
−i[ΣDt]V = µ2ε
∫ddk
(2π)d(iG1σµ)
(imf
(k + p)2 −m2f
)(−iG2σν)
( −ik2 −m2
V
)(gµν +
(ξ − 1)kµkν
k2 − ξm2V
), (6.350)
Using σµσνgµν = D [see eq. (A.8)] and k ·σk ·σ = k2 [from eq. (2.39)] yields
[ΣDt]V = imfG1G2µ2ε
∫ddk
(2π)d1
[(k + p)2 −m2f ][k
2 −m2V ]
[D +
(ξ − 1)k2
k2 − ξm2V
](6.351)
=1
16π2mfG1G2IFV (s;m2
f ,m2V ) (6.352)
where
IFV (s;x, y) ≡ −3 + ξ
ε− 3B(s;x, y)− ξB(s;x, ξy) + 2δMS, (6.353)
after dropping terms that vanish as ε→ 0. It is useful to note that for massless gauge bosons
IFV (x;x, 0) ≡ −3 + ξ
ε+ (3 + ξ)[ln(x/Q2)− 2] + 2δMS. (6.354)
The relevant couplings are obtained from the rules of figs. 67 and 79:
for V = g : G1 = −G2 = gsTa; (6.355)
for V = γ : G1 = −G2 = eQt; (6.356)
for V = Z : G1 = g(T t3 − s2WQt)/cW ; G2 = gs2WQt/cW ; (6.357)
and mf = mt in each case. Adding up these contributions and taking s = m2t , we have:
ΣDt =mt
16π2
g2[(T t3 − s2WQt)s2WQt/c2W
]IFV (m2
t ;m2t ,m
2Z)− (g2
sCF + e2Q2t )IFV (m2
t ;m2t , 0)
+1
2Y 2t IFS(m2
t ;m2t ,m
2hSM
)− 1
2Y 2t IFS(m2
t ;m2t , ξm
2Z)− Y 2
b IFS(m2t ;m
2b , ξm
2W )
. (6.358)
In each of the self-energy functions above, there are poles in 1/ε, contained within the
functions IFV , IFS , IFV and IFS . In the MS or DR schemes, these poles are simply canceled
by counterterms, which have no other effect at one-loop order. The one-loop top-quark pole
mass can now be obtained by plugging eqs. (6.332), (6.342), and (6.358) into eq. (6.312) with
ξ = 0, as discussed earlier. It is not hard to check that the terms from the vector exchange
diagrams that depend on ξm2W and ξm2
Z then just cancel against contributions from massless
Nambu-Goldstone bosons.
As a simple example, consider the one-loop pole mass with only QCD effects included.
Then the result of eq. (6.312) has no imaginary part. Taking the square root (and dropping a
107
two-loop order part) yields the well-known result [119]:
Mt,pole = mt(1 + 12ΣLt + 1
2ΣRt) + ΣDt (6.359)
= mt
(1− CF g
2s
16π2
[IFV (m2
t ;m2t , 0) + IFV (m2
t ;m2t , 0)
])(6.360)
= mt
(1 +
αs4πCF
[5− δMS − 3 ln(m2
t /Q2)]). (6.361)
As another check, consider the imaginary part of the pole squared mass. Equation (6.312)
implies, at leading order:
Γt = −Im[mt(ΣLt + ΣRt) + 2ΣDt] (6.362)
=mt
16π2Im[g2
2IFV (m2
t ;m2b ,m
2W ) + (Y 2
t + Y 2b )IFS(m2
t ;m2b , ξm
2W )
+2Y 2b IFS(m2
t ;m2b , ξm
2W )]. (6.363)
=1
32π2mt
(g2 + Y 2
t + Y 2b )(m2
t +m2b −m2
W )− 4Y 2b m
2t
Im[B(m2
t ;m2b ,m
2W )]. (6.364)
The fact that the ξ dependence cancelled here is a successful check of gauge-fixing invariance,
since the tadpole diagram in fig. 57 does not contribute to the absorptive part of the self-energy.
Now, using
Im[B(s;x, y)] =
0 for s ≤ (
√x+√y)2,
πλ1/2(s, x, y)/s for s > (√x+√y)2.
(6.365)
in eq. (6.364) reproduces the result of eq. (6.11) for the top-quark width at leading order.
6.24 Self-energy and pole mass of the gluino
The Feynman diagrams for the gluino self-energy are shown in Fig. 59. Since the gluino is a
Majorana fermion, we can use the general formalism of subsection 4.6. We will compute the self-
energy functions Ξg ≡ Ξgg and Ωg ≡ Ωgg defined in Fig. 20, and infer Ωg ≡ Ωgg from the latter
by replacing all Lagrangian parameters by their complex conjugates.38 From the general result
of eq. (4.63), it follows that the gluino complex pole squared mass is related to the tree-level
mass mg by
M2g − iMgΓg =
[m2g(1 + 2Ξg) +mg(Ωg + Ωg)
] ∣∣∣s=m2
g+iε(6.366)
at one-loop order.
It is convenient to split the self-energy functions into gluon/gluino loop and squark/gluino
loop contributions, as
Ξg = [Ξg]g +∑
q
∑
x=1,2
[Ξg]qx , and Ωg = [Ωg]g +∑
q
∑
x=1,2
[Ωg]qx , (6.367)
38Suppressing the color degrees of freedom, Ξ, Ω and Ω are one-dimensional matrices, so we do not employboldface letters in this case.
108
p
−ip·σ Ξg
=
g g g
g
+
g q g
qx+
g qc g
qx
p
−iΩg
=
g g g
g
+
g q qc g
qx+
g qc q g
qx
Figure 59: Self-energy functions for the gluino in supersymmetry. The external momentum pµ
flows from the right to the left. The loop momentum kµ in the text is taken to flow clockwise.Spinor and color indices are suppressed. The index x = 1, 2 labels the two squark mass eigen-states of a given flavor q = u, d, s, c, b, t. Both x and q must be summed over. The external legsare amputated.
where the sum over q runs over the six squark flavors u, d, s, c, b, t, and x = 1, 2 corresponds to
the two squark mass eigenstates [i.e., the two appropriate linear combinations (for fixed squark
flavor) of qL and qR]. The gluon exchange contributions, following from the Feynman rules of
Fig. 79, are:
−ip·σ [Ξg]g δab = µ2ε
∫ddk
(2π)d(−gsfaecσµ)
(i(k + p)·σ
(k + p)2 −m2g
)(−gsf ebcσν
)
(−ik2
)(gµν + (ξ − 1)
kµkν
k2
), (6.368)
−i [Ωg]g δab = µ2ε
∫ddk
(2π)d(gsf
eacσµ)
(img
(k + p)2 −m2g
)(−gsf ebcσν
)
(−ik2
)(gµν + (ξ − 1)
kµkν
k2
). (6.369)
The internal gluon and gluino lines carry SU(3)c adjoint representation index indices c and e
respectively, while the external gluinos on the left and right carry indices a and b respectively.
The gluino external momentum pµ flows from right to left, and the loop momentum kµ flows
clockwise. Comparing with the derivations of eqs. (6.330) and (6.352) in the previous subsection,
and using −faecf ebc = f eacf ebc = δabCA [with CA = 3 for SU(3)c], we can immediately conclude
that
[Ξg]g = −αs4πCAIFV (s;m2
g, 0), (6.370)
[Ωg]g = −αs4πCAmgIFV (s;m2
g, 0), (6.371)
where the loop integral functions IFV and IFV were defined in eqs. (6.331) and (6.353).
Next consider the virtual squark-exchange diagrams contributing to Ξ g. Labeling the quark
109
and squark with color indices j, k respectively, we have for each squark mass eigenstate:
−ip·σ [Ξg]qx δab = µ2ε
∫ddk
(2π)d(−i√
2gsTakj Lqx
)( i(k + p)·σ(k + p)2 −m2
q
)(−i√
2gsTbjk L
∗qx
)( i
k2 −m2qx
)
+µ2ε
∫ddk
(2π)d
(i√
2gsTajk R∗
qx
)( i(k + p)·σ(k + p)2 −m2
q
)(i√
2gsTbkj Rqx
)( i
k2 −m2qx
). (6.372)
This uses the Feynman rules shown in fig. 81, given in terms of the squark mixing parameters
Lqx and Rqx defined in eqs. (I.29) and (I.30). Using Tr[T aT b] = 12δab and |Lqx|2 + |Rqx |2 = 1,
and comparing to the derivation of eq. (6.322) of the previous subsection, we obtain:
[Ξg]qx = −αs4πIFS(s;m2
q ,m2qx). (6.373)
Similarly, for the last two diagrams of Fig. 59, we obtain:
−i[Ωg]qx δab = µ2ε
∫ddk
(2π)d
(−i√
2gsTajk L∗
qx
)( imq
(k + p)2 −m2q
)(i√
2gsTbkj Rqx
)( i
k2 −m2qx
)
+µ2ε
∫ddk
(2π)d
(i√
2gsTakj Rqx
)( imq
(k + p)2 −m2q
)(−i√
2gsTbjk L
∗qx
)( i
k2 −m2qx
), (6.374)
again using the Feynman rules shown in fig. 81. As before, j and k are the color indices for the
quark and the squark, respectively. Comparing to the derivation of eq. (6.345) of the previous
subsection, we obtain:
[Ωg]qx = −αs2πL∗qxRqxmqIFS(s;m2
q ,m2qx) . (6.375)
Summing up the results obtained above, and taking s = m2g, we have:
Ξg = −αs4π
[CAIFV (m2
g;m2g, 0) +
∑
q
∑
x=1,2
IFS(m2g;m
2q ,m
2qx)
], (6.376)
Ωg = −αs4π
[CAmgIFV (m2
g;m2g, 0) + 2
∑
q
∑
x=1,2
L∗qxRqxmqIFS(m2
g;m2q ,m
2qx)
]. (6.377)
As previously noted, we can now write down Ωg by replacing the Lagrangian parameters of
eq. (6.377) by their complex conjugates:
Ωg = −αs4π
[CAmgIFV (m2
g;m2g, 0) + 2
∑
q
∑
x=1,2
LqxR∗qxmqIFS(m2
g;m2q ,m
2qx)
]. (6.378)
Inserting the results of eqs. (6.376)–(6.378) into eq. (6.366), one obtains the result [120, 121]:
M2g − iMgΓg = m2
g
[1 +
αs2π
CA[5− δMS − 3 ln
(m2g/Q
2)]
−∑
q
∑
x=1,2
[IFS(m2
g;m2q ,m
2qx) + 2Re[L∗
qxRqx ]mq
mgIFS(m2
g;m2q,m
2qx)]]
, (6.379)
with δMS defined in eq. (6.315).
110
6.25 Triangle anomaly from chiral fermion loops
As our final example, we consider the anomaly in chiral symmetries for fermions, arising from
the triangle diagram involving three currents carrying vector indices.39 Since the anomaly is
independent of the fermion masses, we simplify the computation by setting all fermion masses
to zero. In four-component notation, the treatment of the anomaly requires care because of
the difficulty in defining a consistent and unambiguous γ5 and the epsilon tensor in dimensional
regularization. The same subtleties arise in two-component language, of course, but in a slightly
different form since γ5 does not appear explicitly.
We shall assemble all the ( 12 , 0) [left-handed] two-component fermion fields of the theory
into a (generally reducible) multiplet ψj . For example, the fermions of the Standard Model are:
ψj = (`k , `ck , νk , qi` , q
ci`), where k = 1, 2, 3 and i = 1, 2, . . . , 6 are flavor labels and ` = 1, 2, 3
are color labels [see Table 1]. The two-component spinor indices are suppressed here. Let the
symmetry generators be given by Hermitian matrices T a, so that the ψj transform as:
δψj = iθa(T a)jkψk, (6.380)
for infinitesimal parameters θa. The matrices T a form a reducible representation of a Lie algebra
of the symmetry group. In particular, the T a have a block-diagonal structure, where each block
separately transforms the corresponding field of ψj according to its symmetry transformation
properties. Some or all of these symmetries may be gauged. The Feynman rule for the cor-
responding currents is the same as for external gauge bosons, (as in Fig. 8), and is shown in
Fig. 60.
µ, a
k
j
−i(T a)jk σµ or i(T a)j
k σµ
Figure 60: Feynman rule for the coupling of a current carrying vector index µ and correspond-ing to the symmetry generator T a acting on ( 1
2 , 0) [left-handed] fermions. Spinor indices aresuppressed.
Figure 61 shows the two Feynman diagrams that contribute at one-loop to the three-point
function of the symmetry currents. Applying the Feynman rules of Fig. 2 (with m = 0) for the
propagators and Fig. 60 for the currents, the sum of these diagrams is given by:
39The discussion here parallels that given in ref. [122], section 22.3.
111
µ, a
ν, bρ, c k +A
k − p+Ak + q +A
p+ q
pq
µ, a
ν, bρ, c k +B
k + p+Bk − q +B
p+ q
pq
Figure 61: Triangle Feynman diagrams leading to the chiral fermion anomaly. Fermion spinorand flavor indices are suppressed. The fermion momenta as labeled flow in the arrow directions.
iΓabcµνρ = (−1)
∫d4k
(2π)4Tr
(−iσµT a)
i(k − p+A)·σ(k − p+A)2
(−iσνT b)i(k +A)·σ(k +A)2
(−iσρT c)i(k + q +A)·σ(k + q +A)2
+(−iσµT ) i(k − q +B)·σ(k − q +B)2
(−iσρT c)i(k +B)·σ(k +B)2
(−iσνT b)i(k + p+B)·σ(k + p+B)2
, (6.381)
where the overall factor of (−1) is due to the presence of a closed fermion loop. The trace is
taken over fermion flavor/group and spinor indices, both of which are suppressed. Writing
Tr(T aT bT c) = dabc +i
4fabc , (6.382)
where **********
Although the symmetrized three-point function is ultraviolet finite, the individual loop mo-
mentum integrals are divergent, and must be defined with care. We do not regularize them by
the usual procedure of continuing to d = 4− 2ε dimensions, because the trace over sigma matri-
ces crucially involves the antisymmetric tensor with four indices, brought in by eqs. (A.24) and
(A.25), for which there is no consistent and unambiguous generalization outside of four dimen-
sions. (This is related to the difficulty of defining γ5 in the four-component spinor notation.)
Because the individual integrals are linearly divergent, we must allow for arbitrary constant
four-vectors Aµ and Bµ as offsets for the loop momentum when defining the loop integrations
for the two diagrams [123]. The existence of these vectors corresponds to an ambiguity in the
regulation procedure, which can be fixed to preserve some of the symmetries, as we will see
below.
The persistence of the symmetry in the quantum theory for the currents labeled by µ, a and
ν, b and ρ, c implies the conservation equations
(p+ q)µΓabcµνρ = 0 , −pνΓabcµνρ = 0 , and − qρΓabcµνρ = 0 , (6.383)
112
respectively. Here we are interested in the potentially anomalous part obtained by symmetrizing
over the indices a, b, c:
Aabcµνρ = 16 iΓ
abcµνρ + [five permutations of a, b, c]. (6.384)
Then the contribution of both diagrams involves the same group theory factor, the anomaly
coefficient
dabc = 12Tr[T a, T bT c]. (6.385)
First, consider the result for (p+ q)µAabcµνρ. This can be simplified by rewriting
(p+ q)µ = (k + q +A)µ − (k − p+A)µ, (6.386)
(p+ q)µ = (k + p+B)µ − (k − q +B)µ (6.387)
in the first and second diagram terms, respectively, and then applying the formulas
v ·σ v ·σ = v2, (6.388)
v ·σ v ·σ = v2, (6.389)
which follow from eqs. (A.1), (A.2). After rearranging the terms using the cyclic property of the
trace, we obtain:
(p+ q)µAabcµνρ = dabc Tr[σκσνσλσρ]Xκλ, (6.390)
where the integral is given by:
Xκλ =
∫d4k
(2π)4
[(k − p+A)κ
(k − p+A)2(k +A)λ
(k +A)2− (k + q +A)κ
(k + q +A)2(k +A)λ
(k +A)2
+(k +B)κ
(k +B)2(k − q +B)λ
(k − q +B)2− (k +B)κ
(k +B)2(k + p+B)λ
(k + p+B)2
]. (6.391)
Naively, this integral appears to vanish, because the first term is equal to the negative of the
fourth term after a momentum shift k → k − p + A − B, and the second term is equal to the
negative of the third term after k → k+q+A−B. However, these momentum shifts are not valid
for the individually divergent integrals. Instead, Xκλ can be evaluated by a Wick rotation to
Euclidean space, followed by isolating the terms that contribute for large k2 and are responsible
for the integral not vanishing, and then use of the divergence theorem in four dimensions to
rewrite the integral as one over a three-sphere with radius tending to infinity. The result is:
Xκλ =i
96π2
[gκλ(p+ q)·(A+B) + (A− 2B)κ(p+ q)λ + (p+ q)κ(B − 2A)λ
]. (6.392)
113
Applying eq. (A.24), we get the result for the anomaly in the current labeled by µ, a:
(p+ q)µAabcµνρ =i
48π2dabc
[(p+ q)ν(A+B)ρ + (A+B)ν(p+ q)ρ − gνρ(p+ q)·(A+B)
+3iενρκλ(p+ q)κ(A−B)λ]. (6.393)
Repeating all of the steps starting with eq. (6.386), we similarly obtain:
−pνAabcµνρ = − i
48π2dabc
[pρ(A+B)µ + pµ(A+B)ρ − gµρp·(A+B) + 3iερµκλp
κ(A−B + 2q)λ],
(6.394)
−qρAabcµνρ = − i
48π2dabc
[qµ(A+B)ν + qν(A+B)µ − gµνq ·(A+B) + 3iεµνκλq
κ(A−B − 2p)λ].
(6.395)
[Alternatively, one can simply note that eq. (6.394) follows from eq. (6.393) by making the
replacements µ → ν, ν → ρ, ρ → µ, A → A + q, B → B − q, p → q, and q → −p − q,
while eq. (6.395) follows from eq. (6.393) by making the replacements µ → ρ, ν → µ, ρ → ν,
A→ A− p, B → B + p, p→ −p− q, and q → p.]
From eqs. (6.393)–(6.395), it is clear that unless A + B = 0, all three symmetries will
definitely be anomalous unless dabc = 0. To avoid this, we choose B = −A, with the result:
(p+ q)µAabcµνρ = − 1
8π2dabcενρκλ(p+ q)κAλ, (6.396)
−pνAabcµνρ =1
8π2dabcερµκλp
κ(A+ q)λ, (6.397)
−qρAabcµνρ =1
8π2dabcεµνκλq
κ(A− p)λ. (6.398)
It is still not possible to avoid an anomaly in all three symmetries if dabc 6= 0. If one wants an
anomaly to arise only in the current labeled by µ, a (for example, if the symmetries labeled by
b, c are gauged), one must now choose A = p− q. The standard result follows:
(p+ q)µAabcµνρ =1
4π2dabcενρκλp
κqλ, (6.399)
−pνAabcµνρ = 0, (6.400)
−qρAabcµνρ = 0. (6.401)
In particular, one cannot gauge all three symmetries a, b, c unless dabc = 0.
In writing down eq. (6.381), we chose to use the rules with σ matrices for the current vertices
and σ matrices for the massless fermion propagators. If we had chosen the opposite prescription,
the net effect would have been to obtain the same results as above, but with the opposite sign for
all terms involving the epsilon tensor, due to the sign difference in eqs. (A.24) and (A.25). This
leads to an overall sign ambiguity in the anomaly amplitude for the antisymmetric combination
114
of vector indices. Since the phase of the amplitude is not observable, this does not lead to
a problem. However, when combining the two diagrams, or when including higher loop order
diagrams, a consistent choice must be made. Note that the evaluation of the anomaly above
relied on combining diagrams with a common spinor trace structure in eq. (6.390).
Appendix A: Two-component spinor identities in d 6= 4
When considering a theory regularized by dimensional continuation, one must be careful in
treating cases with contracted spacetime vector indices µ, ν, ρ, . . .. Instead of taking on 4 possible
values, these vector indices formally run over d values, where d is infinitesimally different from
4. This means that some identities that would hold in unregularized 4-dimensional theories are
inconsistent and must not be used; other identities remain valid if d replaces 4 in the appropriate
spots; and still other identities hold without modification.
Two important identities that do hold in d 6= 4 dimensions are:
[σµσν + σνσµ]αβ = 2gµνδβα , (A.1)
[σµσν + σνσµ]αβ = 2gµνδαβ. (A.2)
The trace identities:
Tr[σµσν ] = Tr[σµσν ] = 2gµν (A.3)
then follow. We also note that the spinor index trace identity
Tr[1] = δαα = δαα = 2 (A.4)
continues to hold in dimensional continuation regularization methods.
In contrast, the Fierz identity (written here in three equivalent forms):
σµαασββµ = 2δα
βδβ α (A.5)
σµαασµββ = 2εαβεαβ (A.6)
σµαασββµ = 2εαβεαβ (A.7)
does not have a consistent, unambiguous meaning outside of 4 dimensions. (See for example
refs. [124–126] and references therein.) However, the following identities that are implied by the
115
Fierz identity do consistently generalize to d 6= 4 spacetime dimensions:
[σµσµ]αβ = dδβα (A.8)
[σµσµ]αβ = dδα
β(A.9)
[σµσνσµ]αβ = (2− d)σναβ
(A.10)
[σµσνσµ]αβ = (2− d)σαβν (A.11)
[σµσνσρσµ]αβ = 4gνρδβα − (4− d)[σνσρ]αβ (A.12)
[σµσνσρσµ]αβ = 4gνρδα
β− (4− d)[σνσρ]αβ , (A.13)
[σµσνσρσκσµ]αβ = −2[σκσρσν ]αβ + (4− d)[σνσρσκ]αβ (A.14)
[σµσνσρσκσµ]αβ = −2[σκσρσν ]αβ + (4− d)[σνσρσκ]αβ . (A.15)
Eq. (A.5) is the basis for other Fierz identities that hold in 4 dimensions, which are given in
detail in Appendix A of ref. [38] as well as [37, 42, 43].
Identities that involve the (explicitly and inextricably 4-dimensional) εµνρκ symbol,
σµσνσρ = gµνσρ − gµρσν + gνρσµ − iεµνρκσκ , (A.16)
σµσνσρ = gµνσρ − gµρσν + gνρσµ + iεµνρκσκ , (A.17)
σµνσρ = i2(gνρσµ − gµρσν + iεµνρκσκ) , (A.18)
σµνσρ = i2(gνρσµ − gµρσν − iεµνρκσκ) , (A.19)
σµσνρ = i2(gµνσρ − gµρσν − iεµνρκσκ) , (A.20)
σµσνρ = i2(gµνσρ − gµρσν + iεµνρκσκ) , (A.21)
σµνσρκ = −14(gνρgµκ − gµρgνκ + iεµνρκ) (A.22)
+ i2(gνρσµκ + gµκσνρ − gµρσνκ − gνκσµρ) ,
σµνσρκ = −14(gνρgµκ − gµρgνκ − iεµνρκ) (A.23)
+ i2(gνρσµκ + gµκσνρ − gµρσνκ − gνκσµρ) .
are also only meaningful in exactly four dimensions. This applies as well to the trace identities
which follow from them.40 For example,
Tr[σµσνσρσκ] = 2 (gµνgρκ − gµρgνκ + gµκgνρ + iεµνρκ) , (A.24)
Tr[σµσνσρσκ] = 2 (gµνgρκ − gµρgνκ + gµκgνρ − iεµνρκ) . (A.25)
This could lead to ambiguities in loop computations where it is necessary to perform the com-
putation in d 6= 4 dimensions (until the end of the calculation where the limit d→ 4 is taken).
40This is analogous to the statement that Tr (γ5γµγνγργκ) = −4iεµνρκ [in our convention where ε0123 = +1]is only meaningful in d = 4 dimensions. In two-component notation, the equivalent result is Tr[σµσνσρσκ −σµσνσρσκ] = 4iεµνρκ. In the literature various schemes have been proposed for defining the properties of γ5 ind 6= 4 dimensions [126]. In two-component notation, this would translate into a procedure for dealing with generaltraces involving four or more σ and σ matrices.
116
However, in practice one typically finds that the above expressions appear multiplied by the met-
ric and/or other external tensors (such as four-momenta appropriate to the problem at hand).
In almost all such cases, two of the indices appearing in eqs. (A.24) and (A.25) are symmetrized
which eliminates the εµνρκ term, rendering the resulting expressions unambiguous. Similarly, the
sum of the above trace identities can be assigned an unambiguous meaning in d 6= 4 dimensions:
Tr[σµσνσρσκ] + Tr[σµσνσρσκ] = 4 (gµνgρκ − gµρgνκ + gµκgνρ) . (A.26)
By repeatedly applying the identities given in eqs. (A.1)–(A.3) to eqs. (A.24) and (A.25) in
4 dimensions and eq. (A.26) in d dimensions, and using the cyclic property of the trace, one can
recursively derive trace formulas for products of 6 or more σ and σ matrices.
Appendix B: Explicit forms for the two-component spinor wave
functions
In this Appendix, we construct the explicit forms for the eigenstates of the spin operator 12~σ ·s,
and we examine their properties.
Consider a spin-1/2 fermion in its rest frame and quantize the spin along a fixed axis
specified by the unit vector
s ≡ (sin θ cosφ , sin θ sinφ , cos θ) , (B.1)
with polar angle θ and azimuthal angle φ with respect to a fixed z-axis. The corresponding spin
states will be called fixed-axis spin states. The relevant basis of two-component spinors χs are
eigenstates of 12~σ ·s, i.e.,
12~σ ·sχs = sχs , s = ±1
2 . (B.2)
In order to construct the eigenstates of 12~σ ·s, we first consider the case where s = z. In
this case, we define the eignestates of 12σ
3 to be:
χ1/2(z) =
1
0
, χ−1/2(z) =
0
1
. (B.3)
By convention, we have set an arbitrary overall multiplicative phase factor for each spinor of
eq. (B.3) to unity. We then determine χs(s) from χs(z) by employing the spin-1/2 rotation
operator that corresponds to a rotation from z to s. However, this rotation operator is not
unique. In the convention adopted here, we first perform a rotation by an angle φ about the z-
axis [the corresponding rotation operator is denoted by R(z , φ)]. This has no effect, of course, on
the unit vector z. However, it does result in the modification of χs(z) by an s-dependent phase
factor. We then rotate by an angle θ about an axis n = (− sinφ, cos φ, 0) [the corresponding
117
rotation operator is denoted by R(n , θ)]. As a result of the product of these two rotations [an
explicit expression for R is given in eq. (B.19)], z is rotated into s. Explicitly, s = Rz, where41
R = R(n , θ)R(z , φ) =
cos θ cosφ − sinφ sin θ cosφ
cos θ sinφ cosφ sin θ sinφ
− sin θ 0 cos θ
. (B.4)
Employing the spin-1/2 rotation operator corresponding to R, we can compute χs(s),
χs(s) = exp (−iθn·~σ/2) exp(−iφσ3/2
)χs(z) , n = (− sinφ , cosφ , 0) . (B.5)
Eq. (B.5) yields explicit forms42 for the eigenstates of 12~σ ·s:
χ1/2(s) =
e−iφ/2 cos
θ
2
eiφ/2 sinθ
2
, χ−1/2(s) =
−e−iφ/2 sin
θ
2
eiφ/2 cosθ
2
. (B.6)
These spinors are normalized such that
χ†s(s)χs′(s) = δss′ , (B.7)
and satisfy the following completeness relation:
∑
s
χs(s)χ†s(s) =
1 0
0 1
. (B.8)
The spinors χs(s) and χ−s(s) are connected by the following relation:
χ−s(s) = −2s
0 1
−1 0
χ∗
s(s) . (B.9)
Consider a spin-1/2 fermion with four-momentum pµ = (E , ~p), with E = (|~p|2 + m2)1/2, and
the direction of ~p given by
p = (sin θp cosφp , sin θp sinφp , cos θp) . (B.10)
Using eqs. (2.69), (2.70) and (B.6), one can employ eqs. (3.19)–(3.22) to obtain explicit expres-
sions for the two-component spinor wave functions x(~p, s), y(~p, s), x(~p, s) and y(~p, s).
41In the more common convention in the literature, the factor of R(z , φ) is absent in the definition of R.However, our choice for R is motivated by the simplicity of the explicit form given in eq. (B.4).
42Note that for φ → φ + 2π, χs(s) → −χs(s), which simply reflects that double-valueness of SU(2) transfor-mations. However, as the overall phase of the spinor wave function is arbitrary, we shall restrict 0 ≤ φ < 2π and0 ≤ θ ≤ π in our definition of χs(s).
118
Additional properties of the χs can be derived by introducing an orthonormal set of unit
three-vectors sa that provide a basis for a right-handed coordinate system. Explicitly,
sa·sb = δab , (B.11)
sa× sb = εabcsc . (B.12)
We shall identify:
s = s3 (B.13)
as the quantization axis s used in defining the third component of the spin of the fermion in
its rest frame. The unit vectors s1 and s2 are then chosen such that eqs. (B.11) and (B.12)
are satisfied. To explicitly construct the sa, we begin with the orthonormal set x , y , z, and
rotate each unit vector with the rotation matrix R given in eq. (B.4), so that sa = R(x , y , z).
That is,
s1µ = (cos θ cosφ, cos θ sinφ, − sin θ) ,
s2µ = (− sinφ, cosφ, 0) ,
s3µ = (sin θ cosφ, sin θ sinφ, cos θ) . (B.14)
We can use the sa to extend the defining equation of χs [eq. (B.2)]:
12 ~σ ·saχs′ = 1
2τass′χs , (B.15)
where the τass′ are the matrix elements of the Pauli matrices.43 That is, 12~σ ·(s1 ± is2) serve
as ladder operators that connect the spinor wave functions χ1/2 and χ−1/2. Using eq. (B.7), it
follows that eq. (B.15) is equivalent to:
χ†s ~σ ·saχs′ = τass′ . (B.16)
To prove eq. (B.16), we use eq. (B.5) to obtain:
χ†s(s)~σ ·saχs′(s) = χ†
s(z) eiφσ3/2 eiθn·~σ/2 ~σ ·sa e−iθn·~σ/2 e−iφσ
3/2 χs′(z) . (B.17)
The above result can be simplified by using the following identity:
e−iθn·~σ/2 σj eiθn·~σ/2 = Rij(n , θ)σi , (B.18)
where
Rij(n , θ) = ninj + (δij − ninj) cos θ − εijknk sin θ . (B.19)
43We use the symbol τ rather than σ to emphasize that the indices of the Pauli matrices τ a are spin labels s, s′
and not spinor indices α, α. The first (second) row and column of the τ -matrices correspond to s = 1/2 (−1/2).For example, τ 3
ss′ = 2sδss′ (no sum over s).
119
Thus,
χ†s(s)~σ ·saχs′(s) = χ†
s(z)~σ · [R(z , −φ)R(n , −θ)sa] χs′(z) . (B.20)
Since
R−1 = R(z , −φ)R(n , −θ) , (B.21)
where R is the rotation matrix given in eq. (B.4), and R−1sa = (x , y , z), it follows that
~σ · [R(z , −φ)R(n , −θ)sa] = σa . (B.22)
Consequently, we end up with
χ†s(s)~σ ·saχs′(s) = χ†
s(z)σaχs′(z) ≡ τass′ , (B.23)
which defines the matrix elements of the Pauli matrices, and our proof of eq. (B.16) is complete.
All the results of this Appendix also apply to the helicity spinors χλ, which are defined to
be eigenstates of 12~σ ·p, i.e.,
12~σ ·pχλ(p) = λχλ(p) , λ = ± 1
2 , (B.24)
where p is given by eq. (B.10). Eqs. (B.3)–(B.9) also apply to the two-component helicity spinors
after taking s = p (i.e., identifying θ = θp and φ = φp). In addition, in analogy with the sa, we
can introduce an orthonormal set of unit three-vectors pa such that
p1µ = (cos θp cosφp, cos θp sinφp, − sin θp) ,
p2µ = (− sinφp, cosφp, 0) ,
p3µ = pµ = (sin θp cosφp, sin θ sinφp, cos θp) . (B.25)
In particular, eqs. (B.11)–(B.16) apply as well to the two-component helicity spinors after taking
sa = pa.
The overall phase of the helicity spinor wave function of a fermion is conventional. However,
an ambiguity arises in the case of a pair of fermions in its two-particle rest frame, in which the
corresponding fermion three-momenta are ~p and −~p, respectively. The helicity spinor wave
function of the second fermion depends on the definition of χλ(−p). In the convention of
ref. [17], χλ(−p) is obtained from χλ(z) via a rotation by a polar angle π− θp and an azimuthal
angle φp + π with respect to the z-direction. Using this convention with our definition of the
spinor wave function [eq. (B.5)] yields χλ(−p) = iχ−λ(p). An alternative convention advocated
by Jacob and Wick [127] is to define χλ(−p) by starting with χ−λ(z) and then rotating the
spinor by polar angle θp and azimuthal angle φp. In this case,
χλ(−p) = χ−λ(p) , (B.26)
120
and the extra phase factor is absent. We shall adopt the convention of eq. (B.26) in constructing
helicity amplitudes of processes involving fermions.
Suppose that the two fermions considered above have equal mass. In the center-of-mass
frame, if the four-momentum of one of the fermions is pµ = (E , ~p), then the four-momentum of
the other fermion is
pµ ≡ (E , −~p) . (B.27)
The following numerical identities are then satisfied: σ ·p = σ ·p and σ ·p = σ ·p. However, in
order to maintain covariance with respect to the undotted and dotted spinor indices, we shall
write these identities as:
σαβ ·p = σ0αα(σαβ ·p)σ0
ββ, (B.28)
σαβ ·p = σ0αα(σαβ ·p)σ0ββ . (B.29)
Taking the matrix square root of both sides of eqs. (B.28) and (B.29) removes one of the factors
of σ0 and σ0, respectively. Hence,
xα(−~p,−λ) =√p·σχ−λ(−p) = σ0
√p·σχλ(~p) = σ0
αβyβ(~p, λ) , (B.30)
where we have used eqs. (3.19) and (B.26). In this way, we can derive the eight possible relations
for the helicity spinor wave functions:
xα(−~p,−λ) = σ0αβyβ(~p, λ) , xα(−~p,−λ) = yβ(~p, λ)σ0 βα , (B.31)
yα(−~p,−λ) = σ0αβxβ(~p, λ) , yα(−~p,−λ) = xβ(~p, λ)σ0 βα , (B.32)
xα(−~p,−λ) = σ0 αβ yβ(~p, λ) , xα(−~p,−λ) = yβ(~p, λ)σ0βα , (B.33)
yα(−~p,−λ) = σ0 αβ xβ(~p, λ) , yα(−~p,−λ) = xβ(~p, λ)σ0βα . (B.34)
Appendix C: Path integral treatment of two-component fermion
propagators
In Section 4.2 we derived the two-component fermion propagators in momentum space, which
are the Fourier transforms of the free field expectation values of time-ordered products of two
two-component fermion fields
〈0|Tξα(x)ξβ(y) |0〉FT ≡∫d4w 〈0| Tξα(x)ξβ(y) |0〉 eip·w , w ≡ x− y , (C.1)
where the (translationally invariant) expectation values 〈0| Tξα(x)ξβ(y) |0〉 are functions of the
coordinate difference w ≡ x− y. In Section 4.2, the Fourier transforms of these quantities were
computed by using the free-field expansion obtained from the canonical quantization procedure,
and then evaluating the resulting spin sums. In this Appendix, we provide a derivation of the
121
same result by employing path integral techniques. We follow the analysis given in Appendix C
of ref. [148] (with a few minor changes in notation). For a similar textbook treatment of two-
component fermion propagators see for example ref. [115]. For the analogous treatment of the
four-component fermion propagator, see for example ref. [76].
We first consider the action for a single massive neutral two-component fermion ξα(x),
coupled to an anticommuting two-component fermionic source term Jα(x) [cf. eq. (3.1)]:
S =
∫d4x (L + Jξ + ξ J) =
∫d4x
12
[iξσµ∂µξ + iξσµ∂µξ −m(ξξ + ξξ)
]+ Jξ + ξ J
, (C.2)
where we have split the kinetic energy term symmetrically into two terms. The generating
functional is given by
W [J, J ] = N
∫DξDξ eiS[ξ, ξ, J, J ] , (C.3)
where N is a normalization factor chosen such that W [0, 0] = 1 and DξD ξ is the integration
measure. It is convenient to Fourier-transform the fields ξ(x), ξ(x) and sources J(x), J(x) in
eq. (C.3), and rewrite the action in terms of the corresponding Fourier coefficients ξ(p), ξ(p) , J (p)
and J(p):
ξα(x) =
∫d4p
(2π)4e−ip·xξα(p) , ξα(x) =
∫d4p
(2π)4eip·xξα(p) , (C.4)
Jα(x) =
∫d4p
(2π)4e−ip·xJα(p) , Jα(x) =
∫d4p
(2π)4eip·x J α(p) . (C.5)
Furthermore, we introduce the integral representation of the delta function
δ(4)(x− x′) =
∫d4p
(2π)4e−ip·(x−x
′) . (C.6)
In order to rewrite eq. (C.3) in a more convenient matrix form, we introduce the following
definitions:
Ω(p) ≡
ξα(−p)
ξα(p)
, X(p) ≡
Jα(p)
J α(−p)
, M(p) ≡
p · σαβ −mδαβ
−mδαβ p · σαβ
.
(C.7)
Note thatM is a Hermitian matrix. We can then rewrite the action [eq. (C.2)] in the following
matrix form [after using eqs. (2.47) and (2.48) to write the product of the spinor field and the
source in a symmetrical fashion]:
S =1
2
∫d4p
(2π)4
(Ω†MΩ + Ω†X +X†Ω
). (C.8)
The linear term in the field Ω can be removed by a field redfinition
Ω′ = Ω +M−1X . (C.9)
122
In terms of Ω′, the action now takes the convenient form:
S =1
2
∫d4p
(2π)4
(Ω′†MΩ′ −X†M−1X
), (C.10)
where the inverse of the matrix M is given by
M−1 =1
p2 −m2
p · σαβ mδαβ
mδαβ p · σαβ
. (C.11)
The Jacobian of the field transformation given in eq. (C.9) is unity. Hence, one can insert
the new action, eq. (C.10), in the generating functional, eq. (C.3) to obtain (after dropping the
primes on the two-component fermion fields):
W [J , J ] = N
∫DξDξ exp
i
2
∫d4p
(2π)4
(Ω†MΩ−X†M−1X
)(C.12)
= N
[∫DξDξ exp
i
2Ω†MΩ
]exp
− i
2
∫d4p
(2π)4X†M−1X
(C.13)
= exp
− i
2
∫d4p
(2π)4X†M−1X
, (C.14)
where we have defined the normalization constant N such that W [0, 0] = 1. Inserting the explicit
forms for X andM into eq. (C.14), we obtain
W [J , J ] = exp
−1
2
∫d4p
(2π)4
(Jα(−p)
ip · σαβp2 −m2
J β(−p) + J α(p)ip · σαβp2 −m2
Jβ(p)
+Jα(−p) imδαβ
p2 −m2Jβ(p) + J α(p)
imδαβp2 −m2
J β(−p))
. (C.15)
Using eq. (2.49), it is convenient to rewrite the first two terms of the integrand on the right-hand
side of eq. (C.15) in two different ways:
1
2
∫d4p
(2π)4
[Jα(−p)
ip · σαβp2 −m2
J β(−p) + J α(p)ip · σαβp2 −m2
Jβ(p)
]
=
∫d4p
(2π)4Jα(−p)
ip · σαβp2 −m2
J β(−p) =
∫d4p
(2π)4J α(p)
ip · σαβp2 −m2
Jβ(p) , (C.16)
where we have changed integration variables from p→ −p in relating the two terms above. The
vacuum expectation value of the time-ordered product of two spinor fields in configuation space
is obtained by taking two functional derivatives of the generating functional with respect to the
sources J and J and then setting J = J = 0 at the end of the computation (see, e.g., ref. [76]).
For example,(−i
−→δ
δJα(x1)
)W [J, J ]
(−i
←−δ
δJ β(x2)
)∣∣∣∣∣J=J=0
= N
∫DξDξ ξα(x1)ξβ(x2) exp i
∫d4xL
= 〈0|Tξα(x1)ξβ(x2)|0〉 , (C.17)
123
where the functional derivatives act in the indicated direction (which ensures that no extra
minus signs are generated due to the anticommutativity properties of the sources and their
functional derivatives). To obtain the two-point functions involving the product of two spinor
fields with different combinations of dotted and undotted spinors, it may be more convenient to
write Jξ = ξJ and/or ξ J = J ξ in eq. (C.3). One can then easily verify the following expressions
for the four possible two-point functions:
〈0|Tξα(x1)ξβ(x2)|0〉 =
(−i
−→δ
δJα(x1)
)W [J, J ]
(−i
←−δ
δJ β(x2)
)∣∣∣∣∣J=J=0
, (C.18)
〈0|T ξα(x1)ξβ(x2)|0〉 =
(−i
−→δ
δJα(x1)
)W [J, J ]
(−i
←−δ
δJβ(x2)
)∣∣∣∣∣J=J=0
, (C.19)
〈0|T ξα(x1)ξβ(x2)|0〉 =
(−i
−→δ
δJα(x1)
)W [J, J ]
(−i
←−δ
δJ β(x2)
)∣∣∣∣∣J=J=0
, (C.20)
〈0|Tξα(x1)ξβ(x2)|0〉 =
(−i
−→δ
δJα(x1)
)W [J, J ]
(−i
←−δ
δJβ(x2)
)∣∣∣∣∣J=J=0
. (C.21)
As an example, we provide details for the evaluation of eq. (C.18). Using eqs. (C.15) and
(C.16), we obtain:
〈0|Tξα(x1)ξβ(x2)|0〉 =
−→δ
δJα(x1)
(∫d4p
(2π)4Jα(−p)
ip · σαβp2 −m2
J β(−p)) ←−
δ
δJ β(x2). (C.22)
Using the chain rule for functional differentiation,
δ
δJα(x1)=
∫d4p1
δJβ(−p1)
δJα(x1)
δ
δJβ(−p1)=
∫d4p1 e
−ip1·x1δ
δJα(−p1), (C.23)
δ
δJ β(x2)=
∫d4p2
δ J α(−p2)
δJ β(x2)
δ
δ J α(−p2)=
∫d4p2 e
ip2·x2δ
δ J β(−p2), (C.24)
after using the inverse Fourier transform of eq. (C.5). Applying eqs. (C.23) and (C.24) to
eq. (C.22), we obtain:
〈0|Tξα(x1)ξβ(x2)|0〉 =
∫d4p
(2π)4e−ip·(x1−x2)
ip · σαβp2 −m2
, (C.25)
which is equivalent to eq. (4.1) of Section 4.2. With the same methods applied to eqs. (C.19)–
(C.21), one can easily reproduce the results of eqs. (4.2)–(4.4).
We next consider the action for a single massive Dirac two-component fermion. We shall
work in a basis of fields where the action, including external anticommuting sources, is given by
S[χ, χ, η, η, Jχ, Jχ, Jη, Jη ]=
∫d4x
[iχσµ∂µχ+ iησµ∂µη −m(χη + χη) + Jχχ+ χJχ + Jηη + ηJη
].
(C.26)
124
The techniques are similar to the ones used above. We introduce Fourier coefficients for all the
fields and sources and define
Ωc(p) ≡
ηα(−p)
χα(p)
, Xc(p) ≡
Jηα(p)
J αχ(−p)
. (C.27)
The action functional, eq. (C.26), can then rewritten in matrix form as before (but with no
overall factor of 1/2):
S =
∫d4p
(2π)4
(Ω†cMΩc + Ω†
cXc +X†cΩc
), (C.28)
where M is again given by eq. (C.7). The rest of the calculation goes through as before with
few modifications, and yields the Dirac two-component fermion free-field propagators given in
eqs. (4.5)–(4.8).
125
Appendix D: Matrix decompositions for mass matrix diagonal-ization
In scalar field theory, the diagonalization of the scalar squared-mass matrix M 2 is straightfor-
ward. For a theory of n complex scalar fields, M 2 is an hermitian n × n matrix that can be
diagonalized by a unitary matrix W :
W †M2W = m2 = diag(m21,m
22, . . . ,m
2n) . (D.1)
For a theory of n real scalar fields, M 2 is a real symmetric n×n matrix that can be diagonalized
by an orthogonal matrix Q:
QTM2Q = m2 = diag(m21,m
22, . . . ,m
2n) . (D.2)
In both cases, the eigenvalues, m2k of M2 are real. This is the standard matrix diagonalization
problem that is treated in all elementary linear algebra textbooks.
In spin-1/2 fermion field theory, the diagonalization of the fermion mass matrix, which is
treated in Section 3.2, does not take any of the above forms. In this appendix, we review the
linear algebra theory relevant for the matrix decompositions associated with the charged and
neutral spin-1/2 fermion mass matrix diagonalizations.
D.1 Singular Value Decomposition
The diagonalization of the charged (Dirac) fermion mass matrix requires the singular value
decomposition of an arbitrary complex matrix M .
Theorem: For any complex n× n matrix M , unitary matrices L and R exist such that
LTMR = MD = diag(m1,m2, . . . ,mn), (D.3)
where the mk are real and non-negative. This is called the singular value decomposition of the
matrix M .
In general, the mk are not the eigenvalues of M . Rather, the mk are the singular values
of the general complex matrix M , which are defined to be the non-negative square roots of
the eigenvalues of M †M (or eqivalently of MM †). An equivalent definition of the singular
values can be established as follows. Since M †M is an hermitian non-negative matrix, its
eigenvalues are real and non-negative and its eigenvectors, wk, defined by M †Mwk = m2kwk,
can be chosen to be orthonormal.44 Consider first the eigenvectors corresponding to the positive
eigenvalues of M †M . Then, we define the vectors vk such that Mwk = mkv∗k. It follows that
m2kwk = M †Mwk = mkM
†v∗k, which yields: M †v∗k = mkwk. Note that these equations also
44We define the inner product of two vectors to be 〈v|w〉 ≡ v†w. Then, v and w are orthonormal if 〈v|w〉 = 0.The norm of a vector is defined by ‖v ‖ = 〈v|v〉1/2.
126
imply that MM †v∗k = m2kv
∗k. The orthonormality of the wk implies the orthonormality of the v∗k
(and hence the vk):
δjk = 〈wj |wk〉 =1
mjmk〈M †v∗j |M †v∗k〉 =
1
mjmk〈v∗j |MM †v∗k〉 =
mk
mj〈v∗j |v∗k〉 , (D.4)
which yields 〈v∗j |v∗k〉 = δjk.
If wi is an eigenvector of M †M with zero eigenvalue, then 0 = w†iM
†Mwi = 〈Mwi|Mwi〉,which implies that Mwi = 0. Likewise, if v∗i is an eigenvector of MM † with zero eigenvalue, then
0 = vT
i MM †v∗i = 〈MTvi|MTvi〉∗, which implies that MTvi = 0. Because the eigenvectors of
MM † [M †M ] can be chosen orthonormal, the eigenvectors corresponding to the zero eigenvalues
of M [MT] can be taken to be orthonormal.45 Finally, these eigenvectors are also orthogonal to
the eigenvectors corresponding to the non-zero eigenvalues of MM † [M †M ]. That is,
〈wj |wi〉 =1
mj〈M †v∗j |wi〉 =
1
mj〈v∗j |Mwi〉 = 0 , (D.5)
and similarly 〈vj |vi〉 = 0, where the index i [j] runs over the eigenvectors corresponding to the
zero [non-zero] eigenvalues.
Thus, we can define the singular values of a general complex matrix M to be the simulta-
neous solutions (with real non-negative mk) of:46
Mwk = mkv∗k , vT
kM = mkw†k . (D.6)
The corresponding vk (wk), normalized to have unit norm, are called the left (right) singular
vectors of M .
Proof of the singular value decomposition theorem: Eqs. (D.4) and (D.5) imply
that the left [right] singular vectors can be chosen to be orthonormal. Consequently, the unitary
matrix L [R] can be constructed such that its kth column is given by the left [right] singular
vector vk [wk]. It then follows from eq. (D.6) that:
vT
kMw` = mkδk` , (no sum over k). (D.7)
In matrix form, eq. (D.7) coincides with eq. (D.3), and the singular value decomposition is
established.
The singular values of a complex matrix M are unique (up to ordering), as they correspond
to the eigenvalues of M †M (or equivalently the eigenvalues of MM †). The unitary matrices L
45In general, the multiplicity of zero eigenvalues of M [MT] is not equal to the multiplicity of zero eigenvaluesof M†M [MM†]. However the latter, which is equal to the number of linearly independent eigenvectors of M †M[MM†] with zero eigenvalue, coincides with the number of linearly independent eigenvectors of M [MT] with zeroeigenvalue. Moreover, the number of linearly independent wi coincides with the number of linearly independent vi.
46One can always find a solution to eq. (D.6) such that the mk are real and non-negative. Given a solutionwhere mk is complex, we simply write mk = |mk|eiθ and redefine vk → vke
iθ to remove the phase θ.
127
and R are not unique. The matrix R can be determined directly from eq. (D.3) by computing
M †DMD = M2
D, which yields:
R†M †MR = M2D . (D.8)
That is, R is the unitary matrix that diagonalizes the non-negative definite matrix M †M .
Since the eigenvectors, wk of M †M are orthonormal, each of the wk corresponding to the
non-degenerate eigenvalues of M †M can be multiplied by an arbitrary phase eiθk . The wk
corresponding to a degenerate eigenvalue of M †M can be replaced by any orthonormal lin-
ear combination of the corresponding wk. It follows that within the subspace spanned by the
eigenvectors of M †M corresponding to non-degenerate eigenvalues, R is uniquely determined
up to multiplication on the right by an arbitrary diagonal unitary matrix. Within the sub-
space spanned by the eigenvectors of M †M corresponding to a given degenerate eigenvalue, R
is determined up to multiplication on the right by an arbitrary unitary matrix.
Once R is fixed, L is determined by eq. (D.3):
L = (MT)−1R∗MD . (D.9)
However, if some of the diagonal elements of MD are zero, then L is not uniquely defined.
Writing MD in 2×2 block form such that the upper left block is a diagonal matrix with positive
diagonal elements and the other three blocks are equal to the zero matrix of the appropriate
dimensions, it follows that, MD = MDW , where
W =
W0
, (D.10)
W0 is an arbitrary unitary matrix whose dimension is equal to the number of zeros that appear
in the diagonal elements of of MD, and
and
are respectively the identity matrix and zero
matrix of the appropriate size. Hence, we can multiply both sides of eq. (D.9) on the right by
W , which means that L is only determined up to multiplication on the right by an arbitrary
unitary matrix whose form is given by eq. (D.10).47
D.2 Takagi Diagonalization
The most general neutral spin-1/2 fermion mass matrix is complex and symmetric. To identify
the physical eigenstates, this matrix must be diagonalized. However, the equation that governs
47Of course, one can reverse the above procedure by first determining the unitary matrix L. Eq. (D.3) impliesthat LTMM†L∗ = M2
D, in which case L is determined up to multiplication on the right by an arbitrary [diagonal]unitary matrix within the subspace spanned by the eigenvectors corresponding to the degenerate [non-degenerate]eigenvalues of MM†. Having fixed L, one can obtain R = M−1L∗MD from eq. (D.3). As above, R is onlydetermined up to multiplication on the right by a unitary matrix whose form is given by eq. (D.10).
128
the identification of the physical fermion states is not the standard unitary similarity transfor-
mation. Instead it is a different diagonalization equation that was discovered by Takagi [69],
and rediscovered many times since [70].48
Theorem: For any complex symmetric n × n matrix M , there exists a unitary matrix Ω
such that:
ΩTM Ω = MD = diag(m1,m2, . . . ,mn) , (D.11)
where the mk are real and non–negative. This is the Takagi diagonalization49 of the complex
symmetric matrix M . For a physics context see for example [67].
In general, the mk are not the eigenvalues of M . Rather, the mk are the singular values of
the symmetric matrix M . From eq. (D.11) it follows that:
Ω†M †MΩ = M2D = diag(m2
1,m22, . . . ,m
2n) . (D.12)
If all of the singular values mk are non-degenerate, then one can determine Ω from eq. (D.12).
This is no longer true if some of the singular values are degenerate. For example, if M =(
0 11 0
),
then the singular value 1 is doubly–degenerate, but eq. (D.12) yields Ω†Ω =
2×2, which does
not specify Ω. That is, in the degenerate case, the physical fermion states cannot be determined
by the diagonalization of M †M . Instead, one must make direct use of eq. (D.11). Below, we shall
present a constructive method for determining Ω that is applicable in both the non-degenerate
and the degenerate cases.
Eq. (D.11) can be rewritten as MΩ = Ω∗MD, where the columns of Ω are orthonormal. If
we denote the kth column of Ω by vk, then,
Mvk = mkv∗k , (D.13)
where the mk are the singular values and the vectors vk are normalized to have unit norm.
Following Ref. [138], the vk are called the Takagi vectors of the complex symmetric n × n
matrix M . The Takagi vectors corresponding to non–degenerate non–zero [zero] singular values
are unique up to an overall sign [phase]. Any orthogonal [unitary] linear combination of Takagi
vectors corresponding to a set of degenerate non–zero [zero] singular values is also a Takagi
vector corresponding to the same singular value. Using these results, one can determine the
degree of non–uniqueness of the matrix Ω. For definiteness, we fix an ordering of the diagonal
elements of MD.50 If the singular values of M are distinct, then the matrix Ω is uniquely
48Subsequently, it was recognized in Ref. [128] that the Takagi diagonalization was first established for nonsin-gular complex symmetric matrices by Autonne [129].
49In Ref. [70], eq. (D.11) is called the Takagi factorization of a complex symmetric matrix. We choose to referto this as Takagi diagonalization to emphasize and contrast this with the more standard diagonalization of normalmatrices by a unitary similarity transformation. In particular, not all complex symmetric matrices are diagonal-izable by a similarity transformation, whereas complex symmetric matrices are always Takagi-diagonalizable.
50Permuting the order of the singular values is equivalent to permuting the order of the columns of Ω.
129
determined up to multiplication by a diagonal matrix whose entries are either ±1 (i.e., a diagonal
orthogonal matrix). If there are degeneracies corresponding to non–zero singular values, then
within the degenerate subspace, Ω is unique up to multiplication on the right by an arbitrary
orthogonal matrix. Finally, in the subspace corresponding to zero singular values, Ω is unique
up to multiplication on the right by an arbitrary unitary matrix.
Proof of the Takagi diagonalization. To prove the existence of the Takagi diagonaliza-
tion of a complex symmetric matrix, it is sufficient to provide an algorithm for constructing the
orthonormal Takagi vectors vk that make up the columns of Ω. This is achieved by rewriting
the n×n complex matrix equation Mv = mv∗ [with m real and non–negative] as a 2n× 2n real
matrix equation [130, 131]:
MR
Re v
Im v
≡
ReM − ImM
− ImM −ReM
Re v
Im v
= m
Re v
Im v
, where m ≥ 0 . (D.14)
Since M = MT, the 2n × 2n matrix MR ≡(
ReM − ImM− ImM −ReM
)is a real symmetric matrix.51 In
particular, MR is diagonalizable by a real orthogonal similarity transformation, and its eigen-
values are real. Moreover, if m is an eigenvalue of MR with eigenvector (Re v , Im v), then −mis an eigenvalue of MR with (orthogonal) eigenvector (− Im v , Re v). This observation implies
that MR has an equal number of positive and negative eigenvalues and an even number of zero
eigenvalues.52 Thus, Eq. (D.13) has been converted into an ordinary eigenvalue problem for
a real symmetric matrix. Since m ≥ 0, we solve the eigenvalue problem MRu = mu for the
eigenvectors corresponding to the non–negative eigenvalues.53 It is straightforward to prove
that the total number of linearly independent Takagi vectors is equal to n. Simply note that the
orthogonality of (Re v1 , Im v1) and (− Im v1 , Re v1) with (Re v2 , Im v2) implies that v†1v2 = 0.
Thus, we have derived a constructive method for obtaining the Takagi vectors vk. If there
are degeneracies, one can always choose the vk in the degenerate subspace to be orthonormal.
The Takagi vectors then make up the columns of the matrix Ω in eq. (D.11). A numerical
package for performing the Takagi diagonalization of a complex symmetric matrix has recently
been presented in ref. [132] (see also refs. [138] and [139] for previous numerical approaches to
Takagi diagonalization).
51The 2n × 2n matrix MR is a real representation of the n × n complex matrix M .52Note that (− Im v , Re v) corresponds to replacing vk in Eq. (D.13) by ivk. However, for m < 0 these solutions
are not relevant for Takagi diagonalization (where the mk are by definition non–negative). The case of m = 0 isconsidered in footnote 53.
53For m = 0, the corresponding vectors (Re v , Im v) and (− Im v , Re v) are two linearly independent eigen-vectors of MR; but these yield only one independent Takagi vector v (since v and iv are linearly dependent).
130
D.3 Relation between Takagi diagonalization and the singular value decompo-sition
The Takagi diagonalization is a special case of the singular value decomposition. If the complex
matrix M in eq. (D.3) is symmetric, then the Takagi diagonalization corresponds to Ω = L = R.
In this case, the left singular vectors and the right singular vectors coincide (wk = vk) and are
identified with the Takagi vectors defined in eq. (D.13). Nevertheless, in contrast to the singular
value decomposition, where R can be determined from eq. (D.8) modulo right multiplication
by a [diagonal] unitary matrix in the [non]-degenerate subspace [and L is then determined by
eq. (D.9) modulo multiplication on the right by eq. (D.10)], the matrix Ω cannot be determined
from eq. (D.12) in cases where there is a degeneracy among the singular values, as previously
noted. For example, one possible singular value decomposition of the matrix M =(
0 11 0
)can be
obtained by choosing R = I2 and L = M , in which case MTMI2 = I2. This, of course, is not
a Takagi diagonalization. Since R is only defined modulo the multiplication on the right by an
arbitrary 2 × 2 unitary matrix O, then at least one singular value decomposition exists that is
also a Takagi diagonalization. For the example under consideration, it is not difficult to deduce
the Takagi diagonalization: ΩTMΩ = I2, where
Ω =1√2
1 i
1 −i
O , (D.15)
and O is any 2× 2 orthogonal matrix.
Since the Takagi diagonalization is a special case of the singular value decomposition, it
seems plausible that one can prove the former from the latter. This turns out to be correct; for
completeness, we provide the proof below. Our second proof depends on the following lemma:
Lemma: For any symmetric unitary matrix V , there exists a unitary matrix U such that
V = UTU .
Proof of the Lemma: For any n×n unitary matrix V , there exists an hermitian matrix H
such that V = exp (iH) (this is the polar decomposition of V ). If V = V T then H = HT = H∗
(since H is hermitian). But, any real symmetric matrix can be diagonalized by an orthogonal
transformation. It follows that V can also be diagonalized by an orthogonal transformation.
That is, there exists a real orthogonal matrix Q such that54 QTV Q = diag (eiθ1 , eiθ2 , . . . , eiθn).
Thus, the unitary matrix
U = diag (eiθ1/2 , eiθ2/2 , . . . , eiθn/2)QT (D.16)
satisfies V = UTU and the theorem is proven. Note that U is unique modulo multiplication on
the left by an arbitrary real orthogonal matrix.
54The eigenvalues of any unitary matrix are complex numbers of unit norm, i.e. pure phases.
131
Second Proof of the Takagi diagonalization. Starting from the singular value de-
composition of M , there exist unitary matrices L and R such that M = L∗MDR†, where MD
is the diagonal matrix of singular values. Since M = MT = R∗MDL†, we have two different
singular value decompositions for M . However, as noted below eq. (D.8), R is unique modulo
multiplication on the right by an arbitrary [diagonal] unitary matrix within the [non-]degenerate
subspace. Thus, it follows that a unitary matrix V exists of the latter form such that L = RV .
Moreover, V = V T. This is manifestly true within the non-degenerate subspace where V is
diagonal. Within the degenerate subspace, MD is proportional to the identity matrix so that
L∗R† = R∗L†. Inserting L = RV then yields V T = V . Using the Lemma proved above, there
exists a unitary matrix U such that V = UTU . Hence, in the singular value decomposition of a
symmetric complex matrix, M = L∗MDR†,
L = RUTU , (D.17)
for some unitary matrix U . Moreover, it is straightforward to show that:
MDU∗ = U∗MD . (D.18)
Within the degenerate subspace, eq. (D.18) is trivially true since MD is proportional to the
identity matrix. Within the non-degenerate subspace V is diagonal; hence we may choose
U = UT = V 1/2, so that eq. (D.18) is true since diagonal matrices commute. Using eqs. (D.17)
and (D.18), we can write the singular value decomposition of M as follows
M = L∗MDR† = L∗MD = R∗U †U∗MDR
† = (RUT)∗MDU∗R† = Ω∗MDΩ† , (D.19)
where Ω ≡ RUT is a unitary matrix. Thus the existence of the Takagi diagonalization of an
arbitrary complex symmetric matrix [eq. (D.11)] is once again proven.
Appendix E: Correspondence to four-component spinor notation
E.1 Dirac matrices and four-component spinors
Four-component spinor notation employs four-component Dirac spinor fields and the 4×4 Dirac
gamma matrices, whose defining property is:
γµ, γν = 2gµν . (E.1)
The correspondence between the two-component notation of this paper and the four-component
Dirac spinor notation is most easily exhibited in the basis in which γ5 is diagonal (this is called
the chiral representation). In 2×2 blocks, the gamma matrices are given by:
γµ =
0 σµ
αβ
σµαβ 0
, γ5 ≡ iγ0γ1γ2γ3 =
−δα
β 0
0 δαβ
. (E.2)
132
In addition, we introduce:55
12Σµν ≡ i
4[γµ, γν ] =
σ
µναβ 0
0 σµναβ
. (E.3)
A four component Dirac spinor field, Ψ(x), is made up of two mass-degenerate two-component
spinor fields, χα(x) and ηα(x) as follows:
Ψ(x) ≡
χα(x)
ηα(x)
. (E.4)
We define chiral projections operators PL ≡ 12(1− γ5) and PR ≡ 1
2(1 + γ5) so that
ΨL(x) ≡ PLΨ(x) =
χα(x)
0
, ΨR(x) ≡ PRΨ(x) =
0
ηα(x)
. (E.5)
The free fields can be expanded in a Fourier series; each mode is multiplied by a commuting
spinor wave function as in eq. (3.66). The Dirac conjugate field Ψ and the charge conjugate field
are respectively given by
Ψ(x) ≡ Ψ†A = (ηα(x), χα(x)) , (E.6)
Ψc(x) ≡ CΨT(x) =
ηα(x)
χα(x)
, (E.7)
where the Dirac conjugation matrix A and the charge conjugation matrix C satisfy [140, 141]:
AγµA−1 = 㵆 , C−1γµC = −γµT . (E.8)
It is conventional to impose two additional conditions:
Ψ = A−1Ψ†, (Ψc)c = Ψ . (E.9)
The first of these conditions together with eq. (E.6) is equivalent to the statement that ΨΨ
is hermitian. The second condition corresponds to the statement that the charge conjugation
operator applied twice is equal to the identity operator. Using eqs. (E.8) and (E.9) and the
defining property of the gamma matrices [eq. (E.1)], one can show (independently of the gamma
matrix representation) that the matrices A and C must satisfy:
A† = A , CT = −C , (AC)−1 = (AC)∗ . (E.10)
55In most textbooks, Σµν is called σµν . Here, we use the former symbol so that there is no confusion with thetwo-component definition of σµν given in eq. (2.58).
133
For completeness, we also introduce a matrix B that satisfies [140, 141]:
BγµB−1 = γµT . (E.11)
The matrix B arises in the study of time reversal invariance of the Dirac equation. In the chiral
representation, A, B and C are explicitly given by
A =
0 δαβ
δαβ 0
, B =
ε
αβ 0
0 −εαβ
, C = −γ5B
−1 =
εαβ 0
0 εαβ
. (E.12)
Note the numerical equalities, A = γ0, B = γ1γ3 and C = iγ0γ2, although these identifications
do not respect the structure of the undotted and dotted indices specified in eq. (E.12). In
calculations that involve translations between two-component and four-component notation, the
expressions given in eq. (E.12) should be used. In calculations involving only four-component
notation, there is no harm in using the numerical values for the matrices noted above.
Using eqs. (E.8) and (E.11), the following results are easily derived:
AΓA−1 = ηAΓ
Γ† , ηAΓ
=
+1 , for Γ =
, γµ , γµγ5 , Σµν ,
−1 , for Γ = γ5 , Σµνγ5 ,(E.13)
BΓB−1 = ηBΓ
ΓT , ηBΓ
=
+1 , for Γ =
, γ5 , γ
µ ,
−1 , for Γ = γµγ5 , Σµν , Σµνγ5 ,(E.14)
C−1ΓC = ηCΓ
ΓT , ηCΓ
=
+1 , for Γ =
, γ5 , γ
µγ5 ,
−1 , for Γ = γµ , Σµν , Σµνγ5 ,(E.15)
where
is the 4× 4 identity matrix.
The external two-component spinor momentum space wave functions are related to the
traditional four-component spinors according to:
u(~p, s) =
xα(~p, s)
yα(~p, s)
, u(~p, s) = (yα(~p, s), xα(~p, s)) , (E.16)
v(~p, s) =
yα(~p, s)
xα(~p, s)
, v(~p, s) = (xα(~p, s), yα(~p, s)) , (E.17)
where v(~p, s) = Cu(~p, s)T. The spin quantum number takes on values s = ± 12 , and refers either
to the component of the spin as measured in the rest frame with respect to a fixed axis or to the
helicity (as discussed in Section 3.1). One can check that u and v satisfy the Dirac equations56
(/p−m)u(~p, s) = (/p+m) v(~p, s) = 0 , u(~p, s) (/p−m) = v(~p, s) (/p+m) = 0 , (E.18)
56We use the standard Feynman slash notation: /p ≡ γµpµ and /S ≡ γµSµ.
134
corresponding to eqs. (3.9)–(3.12), and
(2sγ5/S − 1)u(~p, s) = (2sγ5/S − 1) v(~p, s) = 0 , u(~p, s) (2sγ5/S − 1) = v(~p, s) (2sγ5/S − 1) = 0 ,
(E.19)
corresponding to eqs. (3.23)–(3.26), where the spin vector Sµ is defined in eq. (3.15). For massive
fermions, eqs. (3.44)–(3.47) correspond to
u(~p, s)u(~p, s) = 12 (1 + 2sγ5/S) (/p+m) , (E.20)
v(~p, s)v(~p, s) = 12(1 + 2sγ5/S) (/p−m) . (E.21)
To apply the above formulas to the massless case we must employ helicity states, where s is
replaced by the helicity quantum number λ, and Sµ is defined by eq. (3.16). In particular, in the
m → 0 limit, Sµ = pµ/m + O(m/E). Inserting this result in eqs. (E.18) and (E.19), it follows
that the massless helicity spinors are eigenstates of γ5
γ5u(~p, λ) = 2λu(~p, λ) , γ5v(~p, λ) = −2λv(~p, λ) . (E.22)
Applying the same limiting procedure to eqs. (E.20) and (E.21) and using the mass-shell con-
dition (/p/p = p2 = m2), one obtains the helicity projection operators for a massless spin-1/2
particle
u(~p, λ)u(~p, λ) = 12(1 + 2λγ5) /p , (E.23)
v(~p, λ)v(~p, λ) = 12(1− 2λγ5) /p , (E.24)
which correspond to eqs. (3.52)–(3.55). Finally, the spin-sum identities
∑
s
u(~p, s)u(~p, s) = /p+m, (E.25)
∑
s
v(~p, s)v(~p, s) = /p−m, (E.26)
∑
s
u(~p, s)vT(~p, s) = (/p+m)CT , (E.27)
∑
s
uT(~p, s)v(~p, s) = C−1(/p−m) , (E.28)
∑
s
vT(~p, s)u(~p, s) = C−1(/p+m) , (E.29)
∑
s
v(~p, s)uT(~p, s) = (/p−m)CT , (E.30)
correspond to eqs. (3.56)–(3.59).
Bilinear covariants are quantities that are quadratic in the Dirac spinor field which transform
irreducibly as Lorentz tensors. These are constructed from corresponding quantities that are
135
quadratic in the two-component fermion fields. To construct a translation table between the
two-component form and the four-component forms for the bilinear covariants, we first introduce
two Dirac spinor fields [cf. eq. (E.4)]:
Ψ1(x) ≡
χ1(x)
η1(x)
, Ψ2(x) ≡
χ2(x)
η2(x)
, (E.31)
where spinor indices have been suppressed on the two-component fields χi(x) and ηi(x).57 The
following results are then obtained:58
Ψ1PLΨ2 = η1χ2 , (E.32)
Ψ1PRΨ2 = χ1η2 , (E.33)
Ψ1γµPLΨ2 = χ1σ
µχ2 , (E.34)
Ψ1γµPRΨ2 = η1σ
µη2 , (E.35)
Ψ1ΣµνPLΨ2 = 2 η1σ
µνχ2 , (E.36)
Ψ1ΣµνPRΨ2 = 2 χ1σ
µν η2 . (E.37)
Note that eqs. (E.32)–(E.37) apply to both commuting and anti-commuting fermion fields. In
particular, the above results imply that the following relations are satisfied by the (commuting)
u and v spinors:
u(~p1, s1)PLv(~p2, s2) = −u(~p2, s2)PLv(~p1, s1) , (E.38)
u(~p1, s1)PRv(~p2, s2) = −u(~p2, s2)PRv(~p1, s1) , (E.39)
u(~p1, s1)γµPLv(~p2, s2) = u(~p2, s2)γ
µPRv(~p1, s1) , (E.40)
u(~p1, s1)γµPRv(~p2, s2) = u(~p2, s2)γ
µPLv(~p1, s1) . (E.41)
If a bilinear combination of the two-component spinors is given that does not conform
to those listed in eqs. (E.32)–(E.37), then the corresponding four-component expression will
necessarily involve a charge-conjugated four-component spinor. For example, Ψc1PLΨ2 = χ1χ2,
etc. In general, if one replaces Ψj with Ψcj (j = 1 and/or 2) in any of the above results, then in
the corresponding two-component expression one simply interchanges χj ↔ ηj and χj ↔ ηj .
57Here i is a flavor index. In the convention of Section 3.2, the flavor index of an unbarred two-component fieldappears as a lowered index and the flavor index of a barred two-component fermion field appears as a raised index.If one wanted to introduce both raised and lowered indices for four-component fermion fields, one would demandthat the flavor indices of ΨL ≡ PLΨ and ΨR ≡ ΨPL appear as lowered indices, whereas the flavor indices ofΨR ≡ PRΨ and ΨL ≡ ΨPR appear as raised indices. We shall follow this convention for chiral theories. However,such a convention is unwieldy for vector-like interactions. Hence when considering vector-like theories, we shalldepart from this flavor index convention, and employ only lowered flavor indices for the fermion fields.
58It is often useful to apply eq. (2.49) to eqs. (E.35), (E.44) and (E.45) and rewrite η1σµη2 = −η2σ
µη1, wherethe minus sign has been employed for anticommuting spinors.
136
Using eqs. (E.32)–(E.37), it then follows that:
Ψ1Ψ2 = η1χ2 + χ1η2 (E.42)
Ψ1γ5Ψ2 = −η1χ2 + χ1η2 (E.43)
Ψ1γµΨ2 = χ1σ
µχ2 + η1σµη2 (E.44)
Ψ1γµγ5Ψ2 = −χ1σ
µχ2 + η1σµη2 (E.45)
Ψ1ΣµνΨ2 = 2(η1σ
µνχ2 + χ1σµν η2) (E.46)
Ψ1Σµνγ5Ψ2 = 2(−η1σ
µνχ2 + χ1σµν η2) . (E.47)
Note that eqs. (E.44) and (E.45) contain both σµ and σµ [see footnote 58]. In addition, for
anticommuting fermion fields, we may use CT = −C to prove that
ΨciΓΨc
j = ΨjCΓTC−1Ψi = ηCΓ
ΨjΓΨi , (E.48)
where the sign ηCΓ
is given in eq. (E.15).
The results derived above also apply to four-component Majorana fermions, ΨMi, by setting
ηi = χi. However, the extra condition imposed by ΨcMi = ΨMi can yield further restrictions.
For example, eqs. (E.44)–(E.47) imply [after employing eqs. (2.49)–(2.51)] that anticommuting
Majorana four-component fermions satisfy:
ΨMiγµPLΨMj = −ΨMjγ
µPRΨMi , (E.49)
ΨMiΣµνΨMj = −ΨMjΣ
µνΨMi , (E.50)
ΨMiΣµνγ5ΨMj = −ΨMjΣ
µνγ5ΨMi . (E.51)
If we set i = j, we learn that ΨMγµΨM = ΨMΣµνγΨM = ΨMΣµνγ5ΨM = 0.
E.2 Feynman rules for four-component fermions
We now illustrate some basic applications of the the above formalism. First, we consider neutral
and charged fermions interacting with a neutral scalar φ or neutral gauge boson Aµa .59 To obtain
the interactions of the four-component fermion fields, we first identify the neutral two-component
fermion mass-eigenstate neutral fields ξi and the mass-degenerate charged pairs χj and ηj that
combine to form the (mass eigenstate) Dirac fermions. Using eqs. (4.12), (4.17) and (4.20), we
write out the following interaction Lagrangian in two-component form:
Lint = −12(λijξiξj + λij ξ
iξj)φ− (κijχiηj + κijχiηj)φ
−(Ga)ij ξiσµξjA
µa + [(GaR)i
j ηiσµηj − (GaL)ijχiσµχj]A
µa , (E.52)
59Here, charged and neutral refer to some global or local U(1), which in general is orthogonal to the gaugegroup under which Aµ
a transforms.
137
φΨMj
ΨMi
−i(λijPL + λijPR)
φΨj
Ψi
−i(κjiPL + κijPR)
Aaµ ΨMj
ΨMi
−iγµ[(Ga)ijPL − (Ga)jiPR]
Aaµ Ψj
Ψi
−iγµ[(GaL)ijPL + (GaR)j
iPR]
Figure 62: Feynman rules for four-component fermion interactions with neutral bosons
where Ga, GaL and GaR are hermitian matrices, λ is a complex symmetric matrix and κ is an
arbitrary complex matrix, with λij ≡ λ∗ij and κij ≡ κ∗ij By assumption, χ and η have the
opposite charges, while all other fields in eq. (E.52) are neutral. It is now simple to convert this
result into four-component notation:
Lint = −12(λijΨMiPLΨMj + λijΨMiPRΨMj)φ− (κjiΨiPLΨj + κijΨiPRΨj)φ
−[(Ga)i
jΨMiγµPLΨMj + (GaL)ijΨiγµPLΨj + (GaR)j
iΨiγµPRΨj
]Aµa , (E.53)
where ΨMj [Ψj ] are a set of Majorana [Dirac] four-component fermions. It is convenient to use
eq. (E.50) to rewrite the term proportional to (Ga)ij in eq. (E.53) as follows
(Ga)ijΨMiγ
µPLΨMj = 12ΨMiγ
µ[(Ga)i
jPL − (Gξ)jiPR
]ΨMj . (E.54)
Using standard four-component methods, the Feynman rules for the vertices are easily
obtained and displayed in Fig. 62. Note that the arrows on the Dirac fermion lines depict the
flow of the conserved charge. A Majorana fermion is neutral under all conserved charges (and
thus equal to its own anti-particle). Thus an arrow on a Majorana fermion line simply reflects
the structure of the interaction Lagrangian; i.e., ΨM [ΨM ] is represented by an arrow pointing
138
out of [into] the vertex. The arrows are then used for determining the placement of the u and
v spinors in an invariant amplitude.
We next treat the interaction of fermions with charged bosons. Here, we consider a set
of neutral fermion mass-eigenstate fields ξi and a set of charged fermions denoted by pairs of
oppositely charged mass-eigenstate fields χj and ηj. The charged scalar and vector bosons are
complex fields denoted by Φ and W , respectively. We shall only consider the simplest case where
the U(1) charges of Φ, W and χ are assumed to be equal, with η having the opposite charge.
The interaction Lagrangian in two-component form is
Lint = −Φ∗[κij1 χiξj + (κ2)ij ηiξj]− Φ[κij2 ηiξj + (κ1)ijχ
iξj ]
−Wµ[(G1)ijχiσµξj + (G2)i
j ξiσµηj ]−W ∗
µ [(G1)jiξjσµχi + (G2)j
iηjσµξi] , (E.55)
where G1 and G2 are hermitian matrices and κ1 and κ2 are complex symmetric matrices, with
(κn)ij ≡ (κijn )∗ [n = 1, 2]. We may then rewrite this in four-component notation:
Lint = −[(κ2)
ijΨiPLΨMj + (κ1)ijΨiPRΨMj
]Φ
−[(G1)i
jΨiγµPLΨMj − (G2)j
iΨiγµPRΨMj
]Wµ + h.c. (E.56)
There is an equivalent form of the interaction given in eq. (E.56) where L is written in
terms of charge-conjugated fields [after using eq. (E.48)]. Noting that Majorana fermions are
self-conjugate, the Feynman rules for the interactions of neutral and charged fermions with
charged bosons can take two possible forms, as shown in Fig. 63. Here, the direction of an arrow
on a Dirac fermion line is meaningful and indicates the direction of charge flow. However, we are
free to choose either a Ψ or Ψc line to represent a Dirac fermion at any place in a given Feynman
graph.60 Moreover, the structure of the interactions above imply that the arrow directions on
fermion lines flow continuously through the diagram. This requirement then determines the
direction of the arrows on Majorana fermion lines.
Virtual Dirac fermion lines can either correspond to Ψ or Ψc. Here, there is no ambiguity
in the propagator Feynman rule, since for free Dirac fermion fields,
⟨0|T (Ψα(x)Ψβ(y))|0
⟩=⟨0|T (Ψc
α(x)Ψcβ(y))|0
⟩, (E.57)
so that the Feynman rule for the propagator of a Ψ and Ψc line, given in Fig. 64, are identical.
Using eq. (E.2), the four-component fermion propagator Feynman rule can be expressed as a
partitioned matrix of 2× 2 blocks,
p
=α β
=
i
p2 −m2 + iε
mδα
β p·σαβ
p·σαβ mδαβ
, (E.58)
60Since the charge of Ψc is opposite to that of Ψ, the corresponding arrow direction of the two lines are alsopoint in opposite directions.
139
ΦΨMj
Ψi
orΦ
ΨMj
Ψci
−i(κ2ijPL + κ1ijPR)
ΦΨMj
Ψi
orΦ
ΨMj
Ψci
−i(κ1ijPL + κ2ijPR)
WΨMj
Ψi
−iγµ(G1ijPL −G2j
iPR)
WΨMj
Ψci
iγµ(G1ijPR −G2j
iPL)
WΨMj
Ψi
−iγµ(G1jiPL −G2i
jPR)
WΨMj
Ψci
iγµ(G1ijPR −G2j
iPL)
Figure 63: Feynman rules for four-component fermion interactions with charged bosons. Thearrows on the boson and Dirac fermion lines indicate the direction of charge flow.
140
p
αβ
i(/p+m)αβp2 −m2 + iε
Figure 64: Feynman rule for propagator of a four-component fermion with mass m. This samerule applies to a Majorana, Dirac and charge-conjugated Dirac fermion. The four-componentspinor labels are specified.
consisting of two-component fermion propagators defined in Fig. 2, with the undotted and dotted
α [β] indices on the left [right] and with the momentum flowing from right to left.
The derivation of the four-component Dirac fermion propagator is treated in most modern
textbooks of quantum field theory [see, e.g., ref. [76]]. Here, we briefly sketch the path-integral
derivation of the four-component fermion propagator by exploiting the path integral treatment
of the two-component fermion propagators outlined in Appendix C. Consider a single massive
Dirac fermion Ψ(x) coupled to an anticommuting four-component Dirac fermionic source term
Jψ(x) ≡
Jηα(x)
J αχ (x)
. (E.59)
The corresponding action [eq. (C.2)] in four-component notation is given by
S =
∫d4x (L + JψΨ + ΨJψ) =
∫d4x
[Ψ(i/∂ −m)Ψ + JψΨ + Ψ Jψ
]. (E.60)
Introducing the momentum space Fourier coefficients:
Ψ(x) =
∫d4p
(2π)4e−ip·xΨ(p) , Jψ(x) =
∫d4p
(2π)4e−ip·xJψ(p) , (E.61)
we can identify the following four-component quantities with matrices of two-component quan-
tities given in eqs. (C.7) and (C.27):
Ψ(p) = A−1Ωc(p) , Jψ(p) = Xc(p) , /p−m =M(p)A , (E.62)
where A is the Dirac conjugation matrix defined in eqs. (E.6) and (E.8). Using the results of
Appendix C, one easily derives:
〈0|T (Ψ(x1)Ψ(x2))|0〉 =
(−i
−→δ
δJψ(x1)
)W [J, J ]
(−i
←−δ
δJψ(x2)
)∣∣∣∣∣Jψ=Jψ=0
, (E.63)
where
W [Jψ , Jψ] = exp
−i∫
d4p
(2π)4Jψ(p)
/p+m
p2 −m2Jψ(p)
. (E.64)
Using the analogs of eqs. (C.23) and (C.24), we end up with the expected result
〈0|T (Ψ(x1)Ψ(x2))|0〉 =
∫d4p
(2π)4e−ip·(x1−x2) /p−m
p2 −m2. (E.65)
141
In principle, the analogous computation can be carried out for a single four-component Ma-
jorana fermion field ΨM (x) coupled to a Majorana fermionic source, Jξ(x). The corresponding
action is similar to that of eq. (E.60), with an extra overall factor of 1/2. However, in evaluating
the functional derivative in eq. (E.64), one must take into account that the Majorana fermionic
source Jξ(x) satisfies J cξ ≡ CJ T
ξ = Jξ. Consequently, the functional derivative with respect to
Jξ is related to the corresponding functional derivative with respect to Jξ. As a result, the cal-
culation of eq. (E.64) will yield two equal terms that will cancel the overall factor of 1/2, and the
end result will again be eq. (E.65). Neverthless, this computation is somewhat awkward using
four-component spinor notation, in contrast to the straightforward calculation of Appendix C.
E.3 Applications of four-component spinor Feynman rules
For a given process, there may be a number of distinct choices for the arrow directions on the
Majorana fermion lines, which may depend on whether one represents a given Dirac fermion by
Ψ or Ψc. However, different choices do not lead to independent Feynman diagrams.61 When
computing an invariant amplitude, one first writes down the relevant Feynman diagrams with
no arrows on any Majorana fermion line. The number of distinct graphs contributing to the
process is then determined. Finally, one makes some choice for how to distribute the arrows
on the Majorana fermion lines and how to label Dirac fermion lines (either as the field or its
conjugate) in a manner consistent with the rules of Figs. 62 and 63. The end result for the
invariant amplitude (apart from an overall unobservable phase) does not depend on the choices
made for the direction of the fermion arrows.
Using the above procedure, the Feynman rules for the external fermion wave functions are
the same for Dirac and Majorana fermions:
• u(~p, s): incoming Ψ [or Ψc] with momentum ~p parallel to the arrow direction,
• u(~p, s): outgoing Ψ [or Ψc] with momentum ~p parallel to the arrow direction,
• v(~p, s): outgoing Ψ [or Ψc] with momentum ~p anti-parallel to the arrow direction,
• v(~p, s): incoming Ψ [or Ψc] with momentum ~p anti-parallel to the arrow direction.
The proof that the above rules for external wave functions apply unambiguously to Majorana
fermions is straightforward. Simply insert the plane wave expansion of the Majorana field:
ΨM (x) =∑
s
∫d3~p
(2π)3/2(2Ep)1/2
[u(~p, s)a(~p, s)e−ip·x + v(~p, s)a†(~p, s)e+ip·x
](E.66)
61In contrast, the two-component Feynman rules developed in Section 3 require that two vertices differing bythe direction of the arrows on the two-component fermion lines must both be included in the calculation of thematrix element.
142
into eq. (E.53), and evaluate matrix elements for, e.g., the decay of a scalar or vector particle
into a pair of Majorana fermions.
We now reconsider the matrix elements for scalar and vector particle decays into fermion
pairs and 2 → 2 elastic scattering of a fermion off a scalar and vector boson, respectively. We
shall compute the matrix elements using the Feynman rules of Fig. 62, and check that the results
agree with the ones obtained by two-component methods in Section 3.
The matrix element for the decay φ→ ΨM (~p1, s1)ΨM (~p2, s2) is given by
iM = −iu(~p1, s1)(λPL + λ∗PR)v(~p2, s2) . (E.67)
One can easily check that this result matches with eq. (4.26), which was derived using two-
component techniques. Note that if one had chosen to switch the two final states (equivalent
to switching the directions of the Majorana fermion arrows), then the resulting matrix element
would simply exhibit an overall sign change [due to the results of eqs. (E.38) and (E.39)].62
Similarly, for φ→ ΨMiΨMj (i 6= j) or for the decay into a pair of Dirac fermions, φ→ ΨΨ, one
again obtains the invariant matrix element given in eq. (E.67).
For the decay Aµ → ΨM (~p1, s1)ΨM (~p2, s2), one obtains:
iM = iGξ u(~p1, s1)γµγ5v(~p2, s2)εµ . (E.68)
One can easily check that this result matches with eq. (4.29). For the decay into non-identical
Majorana fermions, Aµ → ΨMiΨMj (i 6= j), we can use the Feynman rules of Fig. 62 to obtain:
iM = −iu(~pi, si)γµ[(Gξ)i
jPL − (Gξ)jiPR
]v(~pj , sj)εµ , (E.69)
Again, we note that if one had chosen to switch the two final states (equivalent to switching
the directions of the Majorana fermion arrows), then the resulting matrix element would simply
exhibit an overall sign change [due to the results of eqs. (E.40) and (E.41)]. Finally, for the decay
of the vector particle into a Dirac fermion-antifermion pair, Aµ → ΨΨ, the matrix element is
given by:
iM = −iu(~p1, s1)γµ(GLPL −GRPR)v(~p2, s2)εµ , (E.70)
which matches the result of eq. (4.33).
Turning to the elastic scattering of a neutral Majorana fermion and a neutral scalar, we
shall examine two equivalent ways for computing the amplitude. Following the rules previously
stated, there are two possible choices for the direction of arrows on the Majorana fermion lines.
Thus, may evaluate either one of the following two diagrams:
62The overall sign change is a consequence of the Fermi-Dirac statistics, and corresponds to changing whichorder one uses to construct the two particle final state.
143
p −p
plus a second diagram in each case (not shown) where the initial and final state scalars are
crossed. Evaluating the first diagram above, the matrix element for φΨM → φΨM is given by:
iM =−i
s−m2u(~p2, s2)(λPL + λ∗PR)(/p+m)(λPL + λ∗PR)u(~p1, s1) + (crossed)
=−i
s−m2u(~p2, s2)
[|λ|2/p+
(λ2PL + (λ∗)2PR
)m]u(~p1, s1) + (crossed) , (E.71)
where m is the Majorana fermion mass, s is the center-of-mass energy squared. Using eqs. (E.2)
and (E.16), one recovers the results of eq. (4.34). Had we chose to evaluate the second diagram
instead, the resulting amplitude would have been given by:
iM =−i
s−m2v(~p1, s1)
[−|λ|2/p+
(λ2PL + (λ∗)2PR
)m]v(~p2, s2) + (crossed) . (E.72)
Using eqs. (E.16) and (E.17) and the results of eqs. (2.47)–(2.49) one can derive the following
results:
v(~p1, s1)v(~p2, s2) = −u(~p2, s2)u(~p1, s1) , v(~p1, s1)γµv(~p2, s2) = u(~p2, s2)γ
µu(~p1, s1) .
(E.73)
Consequently, the amplitude computed in eq. (E.72) is just the negative of eq. (E.71). This is
expected, since the order of spinor wave functions (12) in eq. (E.72) is an odd permutation (21)
of the order of spinor wave functions in eq. (E.71). As in the two-component Feynman rules,
the overall sign of the amplitude is arbitrary, but the relative signs of any pair of diagrams is
not ambiguous. This relative sign is positive [negative] if the permutation of the order of spinor
wave functions of one diagram relative to the other diagram is even [odd].
Next, we consider the elastic scattering of a charged fermion and a neutral scalar. Again,
we examine two equivalent ways for computing the amplitude. Following the rules previously
stated, there are two possible choices for the direction of arrows on the fermion lines, depending
on whether we represent the fermion by Ψ or Ψc. Thus, we may evaluate either one of the
following two diagrams:
p
Ψ Ψ
−p
Ψc Ψc
plus a second diagram in each case (not shown) where the initial and final state scalars are
crossed. Evaluating the first diagram above, the matrix element for φΨ → φΨ is given by
144
eq. (E.71), with λ replaced by κ. Had we chose to evaluate the second diagram instead, the
resulting amplitude would have been given by eq. (E.72), with λ replaced by κ. Thus, the
discussion above in the case of neutral fermion scattering processes also applies to charged
fermion scattering processes.
In processes that only involve vertices with two Dirac fields, it is never necessary to use
charge-conjugated Dirac fermion lines. In contrast, consider the following process that involves
a vertex with one Dirac and one Majorana fermion. Specifically, we examine the scattering of
a charged Dirac fermion and a charged scalar via the exchange of a neutral Majorana fermion,
in which the charge of the outgoing fermion is opposite to that of the incoming fermion. If
one attempts to draw the relevant Feynman diagram employing Dirac fermion lines but with
no charge-conjugated Dirac fermion lines, one finds that there is no possible choice of arrow
direction for the Majorana fermion that is consistent with the the vertex rules of Fig. 63. The
resolution is simple: one can choose the incoming line to be Ψ and the outgoing line to be Ψc
or vice versa. Thus, the two possible choices are given by:
p
Ψ Ψc
−p
Ψc Ψ
plus a second diagram in each case (not shown) in which the initial and final scalars are crossed.
If we evaluate the first diagram, the resulting amplitude is given by:
iM =−i
s−m2u(~p2, s2)(κ1PL + κ∗2PR)(/p+m)(κ1PL + κ∗2PR)u(~p1, s1) + (crossed)
=−i
s−m2u(~p2, s2)
[κ1κ
∗2/p+
(κ2
1PL + (κ∗2)2PR
)m]u(~p1, s1) + (crossed) , (E.74)
where m is the Majorana fermion mass. One can check that this is equivalent to eq. (4.38)
obtained via the two-component methods. Had we evaluated the second diagram, then after
using the relations given in eq. (E.73), one finds that the resulting amplitude is just the negative
of eq. (E.74), as expected. As before, the relative sign between diagrams for the same process
is not ambiguous.
In the literature, there are a number of alternative methods for dealing with scattering
processes involving Majorana particles. For example, one can define a fermion-number violating
propagator for four-component fermions (see, e.g., [48]). These methods involve subtle choices
of signs which often require first-principles computations to verify. The advantage of the method
described above is that there is never any ambiguity in the choice of relative signs.
In the case of elastic scattering of a fermion and a neutral vector boson, the two contributing
diagrams are
145
p
plus a second diagram (not shown) where the initial and final state vector bosons are crossed.
Consider first the scattering of a neutral Majorana fermion of mass m. Using the Feynman rules
of Fig. 62, we see that the Feynman rule for the AµΨMΨM vertex is given by iGξγµγ5. Hence,
the corresponding matrix element is given by
iM =−iG2
ξ
s−m2u(~p2, s2) γ ·ε∗2 (/p−m) γ ·ε1u(~p1, s1) + (crossed) , (E.75)
where we have used γνγ5(/p +m)γµγ5 = γν(/p −m)γµ. Using eqs. (E.2) and (E.16), one easily
recovers the results of eq. (4.35).
Next, the invariant matrix element for the scattering of a Dirac fermion is given by
iM =−i
s−m2u(~p2, s2) γ ·ε∗2 (GLPL −GRPR)(/p+m) γ ·ε1 (GLPL −GRPR)u(~p1, s1) + (crossed)
=−i
s−m2u(~p2, s2) γ ·ε∗2
[(G2
LPL +G2RPR)/p−GLGRm
]γ ·ε1u(~p1, s1) + (crossed) . (E.76)
One can easily check that this result coincides with that of eq. (4.39).
Finally, we examine the elastic scattering of two identical Majorana fermions via scalar
exchange. The three contributing diagrams are:
and the corresponding matrix element is given by
iM =−i
s−m2φ
[v1(λPL + λ∗PR)u2 u3(λPL + λ∗PR)v4]
+ (−1)−i
t−m2φ
[u3(λPL + λ∗PR)u1 u4(λPL + λ∗PR)u2]
+−i
u−m2φ
[u4(λPL + λ∗PR)u1 u3(λPL + λ∗PR)u2] , (E.77)
where ui ≡ u(~pi, si), vj ≡ u(~pj, sj) and mφ is the exchanged scalar mass. The relative minus
sign of the t-channel graph relative to the other two is obtained by noting that 3142 [4132] is an
146
odd [even] permutation of 1234. Using eqs. (E.2) and (E.16), one easily recovers the results of
eq. (4.40).
subsection*E.4 Self-energy functions and pole masses for four-component fermions
In this section, we examine the self-energy functions and the pole masses for a set of four-
component fermions. We first consider four-component Dirac fermion fields Ψαi, where α is
the four-component spinor index and i is the flavor index. The full, loop-corrected Feynman
propagators with four-momentum pµ are defined by the Fourier transforms [cf. footnote 13] of
vacuum expectation values of time-ordered products of bilinears of the fully interacting four-
component fermion fields:
〈0| TΨαi(x)Ψβj(y) |0〉FT = i(Sαβ)ij(p) , (E.78)
with [149–159]
S(p) ≡ /p[PLST
L(p2) + PRSR(p2)]
+ PLST
D(p2) + PRSD(p2) , (E.79)
where the four-component spinor indices α and β and the flavor indices i and j have been
suppressed. As in Section 4.6, we shall organize the computation of the full propagator in terms
of the 1PI self-energy function [152]:63
Σ(p) ≡ /p[PLΣL(p2) + PRΣT
R(p2)]
+ PLΣD(p2) + PRΣT
D(p2) . (E.80)
Diagrammatically, iS and −iΣ are shown in Fig. 65.
βα
ji
p
i(Sαβ)ij(p)
p
βα
i j
−i(Σαβ)ij(p)
Figure 65: The full, loop-corrected propagator for four-component Dirac fermions, i(Sαβ)ij(p),is denoted by the shaded box, which represents the sum of all connected Feynman diagrams,with external legs included. The self-energy function for four-component Dirac fermions,−i(Σαβ)ij(p), is denoted by the shaded circle, which represents the sum of all one-particleirreducible, connected Feynman diagrams with the external legs amputated. In both cases, Thefour-momentum p flows from right to left.
The hermiticity of the effective action implies that S and Σ satisfy hermiticity condi-
tions [142, 143]
[ST]? = ASA−1 , [ΣT]? = AΣA−1 , (E.81)
63Our notation in eq. (E.80) differs from that of ref. [152], as we employ ΣT
R instead of ΣR. Our motivation forthis choice is that in the case of Majorana fermions [cf. eq. (E.92)], we simply have ΣL = ΣR, without an extratranspose (or conjugation). We have also chosen to employ ST
L in eq. (E.79) for similar reasons.
147
where A is the Dirac conjugation matrix (A = γ0 in all common representations) and the star
symbol was defined in the paragraph below eq. (4.47). Applying eq. (E.81) to eqs. (E.79) and
(E.80) then yields the following conditions for the complex matrix functions:
[ST
L]? = SL , [ST
R]? = SR , SD = S ?D , (E.82)
[ΣT
L]? = ΣL , [ΣT
R]? = ΣR , ΣD = Σ ?D . (E.83)
Starting at tree-level and comparing with Fig. 64, the full propagator function is given by:
Sij(p) = (/p+m)δij/(p2 −m2
i ) + . . . , (E.84)
with no sum over i implied. The full loop-corrected propagator can be expressed diagrammati-
cally in terms of the 1PI self-energy function:
βα
ji=
βα
ji
γα
ki
δ
`
β
j+
(E.85)
As in Section 4.6, the algebraic representation of eq. (E.85) can be written as
S = T + TΣS = (T −1 −Σ)−1 , (E.86)
where T ij ≡ (/p + m)δij/(p2 −m2
i ) is the tree-level contribution to S given in eq. (E.84). By
writing the expressions for S and Σ given in eqs. (E.79) and (E.80) and T in block matrix form
using eq. (E.2), one can verify that eq. (E.86) is equivalent to eq. (4.66). Consequently, the
complex pole masses of the corresponding Dirac fermions are again determined from eq. (4.71).
In the special case of a parity-conserving vectorlike theory of Dirac fermions (such as QED
or QCD), the pseudoscalar and pseudovector parts of S(p) and Σ(p) must be absent. Thus,
the following relations must hold among the loop-corrected propagator functions and self-energy
functions, respectively:
SR = ST
L , SD = [S T
D]? , (E.87)
ΣL = ΣT
R , ΣD = [ΣT
D]? , (E.88)
in agreement with eqs. (4.72) and (4.73).
In the case of a set of four-component Majorana fermion fields, we can still use the results of
eqs. (E.79)–(E.86). However, one obtains additional constraints on the full propagator and self-
energy matrix functions due to the Majorana condition ΨMi = CΨT
Mi. Inserting this result into
eq. (E.78), and making use of the anti-commutativity of the fermion fields, one easily derives:
〈0| TΨMαi(x)ΨMβj(y) |0〉FT = Cαγ 〈0| TΨMδi(x)ΨMγj(y) |0〉FT C−1δβ . (E.89)
148
Consequently,
CSTC−1 = S , CΣTC−1 = Σ . (E.90)
Inserting the expressions for S and Σ [eqs. (E.79) and (E.80)] and using the result of eq. (E.48),
it follows that:
SL = SR , SD = ST
D , SD = ST
D , (E.91)
ΣL = ΣR , ΣD = ΣT
D , ΣD = ΣT
D . (E.92)
As expected, with these constraints the form of eq. (4.66) matches precisely with the form
of eq. (4.56), corresponding to the equation for the full propagator functions for a theory of
generic two-component fermion fields. In the notation of Section 4.6, we can therefore identify:
C ≡ SL = SR , D ≡ SD , Ξ ≡ ΣL = ΣR, and Ω ≡ ΣD.
Appendix F: Covariant spin operators and the Bouchiat-Michel
Formulae
Bouchiat and Michel derived a useful set of formulae [135] that generalize the spin-projection
opertors used in four-component spinor computations. In this Appendix, we establish the two-
component analogues of the Bouchiat-Michel formulae, and demonstrate their equivalence to
the corresponding four-component spinor formulae.
F.1 The covariant spin operators for a spin-1/2 fermion
Consider a massive spin-1/2 fermion of mass m and four-momentum p. We define a set of three
four-vectors Saµ (a = 1, 2, 3) such that the Sa and p/m form an orthonormal set of four-vectors.
In the rest frame of the fermion, where pµ = (m ; ~0), we can define
Saµ ≡ (0 ; sa) , a = 1, 2, 3 , (F.1)
where the unit vectors sa are a mutually orthonormal set of unit three-vectors that form a basis
for a right-handed coordinate system. Explicit forms for the sa are given in eq. (B.14). Using
eq. (2.71), the three four-vectors Saµ in a reference frame in which the four momentum of the
fermion is pµ = (E ; ~p) is given by:
Saµ =
(~p·sa
m; sa +
(~p·sa) ~p
m(E +m)
). (F.2)
As discussed in Appendix B, we identify s = s3 as the quantization axis used in defining the
third component of the spin of the fermion in its rest frame. It then follows that the spin
four-vector, previously introduced in eq. (3.15) is given by Sµ = S3µ.
149
The orthonormal set of four four-vectors p/m and the Sa satisfy the following Lorentz-
covariant relations:
p·Sa = 0 , (F.3)
Sa ·Sb = −δab , (F.4)
εµνλσpµS1νS
2λS
3σ = −m, (F.5)
SaµSbν − SaνSbµ = εabcεµνρσ(S
c)ρpσ
m, (F.6)
Saµ Saν = −gµν +
pµpνm2
, (F.7)
where the sum over the repeated indices is implicitly assumed. It is convenient to define:
Sss′ ≡ Saτass′ , s, s′ = ±12 , (F.8)
where τass′ are the matrix elements of the Pauli matrices (see footnote 43). Then, we can rewrite
eqs. (F.4) and (F.6) as:
gµνSµSν = −3 , (F.9)
SµSν − SνSµ =i
2mεµνρσS
ρpσ . (F.10)
The Sµ serve as covariant spin operaotrs for a spin-1/2 fermion. In particular, in the rest frame,
the 12S
i satisfy the usual SU(2) commutation relations, with ~S2 = 3
4 as expected for a spin-1/2
particle.
It is often desirable to work with helicity states. In this case, we choose:
sa = pa , i.e., θ = θp and φ = φp , (F.11)
where the pa are defined in eq. (B.25) [with p3 ≡ p], in which case eqs. (F.1)–(F.7) also apply
to the two-component helicity spinors. Moreover, since pa ·p = 0 for a 6= 3, it follows that
Sa = (0 ; pa) for a = 1, 2 in all reference frames obtained from the rest frame by a boost in
the p direction. Hence, in a reference frame where pµ = (E ; ~p), eqs. (B.14) and (F.2) provide
explicit forms for the Sa,
S1µ = (0 ; cos θ cosφ, cos θ sinφ, − sin θ) , (F.12)
S2µ = (0 ; − sinφ, cosφ, 0) , (F.13)
S3µ =
( |~p |m
;E
mp
), (F.14)
in a coordinate system where p = (sin θ cosφ, sin θ sinφ, cos θ). As expected, S 3µ is the spin
vector for helicity states obtained in eq. (3.16). In the high energy limit (E m),
mSaµ = pµ δa3 +O(m) . (F.15)
150
F.2 Two-component spinor wave function relations
In Section 3.1, we wrote down explicit forms for the undotted spinor wave functions
xα(~p, s) =√p·σ χs , xα(~p, s) = −2sχ†
−s√p·σ , (F.16)
yα(~p, s) = 2s√p·σ χ−s , yα(~p, s) = χ†
s
√p·σ , (F.17)
and the dotted spinor wave functions
xα(~p, s) = −2s√p·σ χ−s , xα(~p, s) = χ†
s
√p·σ , (F.18)
yα(~p, s) =√p·σ χs , yα(~p, s) = 2sχ†
−s√p·σ , (F.19)
where√p·σ and
√p·σ are defined in eqs. (2.69) and (2.70). As shown in Appendix B, the
two-component spinors χs satisfy:
12 ~σ ·saχs′ = 1
2τass′χs , χ†
s(s)χs′(s) = δss′ , s , s′ = ±12 . (F.20)
Next, we use eqs. (2.72) and (2.73) to obtain:
√p·σ Sa ·σ√p·σ = m~σ ·sa , (F.21)
√p·σ Sa ·σ
√p·σ = −m~σ ·sa , (F.22)
which extends the results of eqs. (3.17) and (3.18). As a result, we obtain a generalization of
eqs. (3.23)–(3.26):
(Sa ·σ)αβxβ(~p, s′) = τa
ss′yα(~p, s) , (Sa ·σ)αβ y
β(~p, s′) = −τass′xα(~p, s) , (F.23)
(Sa ·σ)αβ xβ(~p, s′) = −τa
s′syα(~p, s) , (Sa ·σ)αβyβ(~p, s
′) = τas′sxα(~p, s) , (F.24)
xα(~p, s′)(Sa ·σ)αβ = −τas′syβ(~p, s) , yα(~p, s
′)(Sa ·σ)αβ = τas′sxβ(~p, s) , (F.25)
xα(~p, s′)(Sa ·σ)αβ = τa
ss′yβ(~p, s) , yα(~p, s′)(Sa ·σ)αβ = −τa
ss′xβ(~p, s) , (F.26)
where there is an implicit sum over the repeated label s = ± 12 . As expected, the case of a = 3
simply reproduces the results of eqs. (3.23)–(3.26) obtained previously. The above equations also
apply to helicity wave functions x(~p, λ) and y(~p, λ) by replacing s, s′ with λ, λ′ and defining
the Sa by eqs. (F.12)–(F.14).
The derivation of eqs. (F.23)–(F.26) for arbitrary a closely follows the corresponding deriva-
tion for a = 3 previously given. For example, using eqs. (F.21) and (F.22) and the definitions
for xα(~p, s) and yα(~p, s), we find (suppressing spinor indices),
√p·σ Sa ·σ x(~p, s′) =
√p·σ Sa ·σ√p·σ χs′ = m~σ ·sa χs′ = mτass′ χs , (F.27)
after using eq. (F.20). Multiplying both sides of eq. (F.27) by√p·σ, we end up with
S ·σ x(~p, s′) = τass′√p·σ χs = τass′ y(~p, s) . (F.28)
151
Similarly,
S ·σx(~p, s′) = 2s′τa−s,−s′√p·σ χ−s = −τas′s y(~p, s) , (F.29)
where we have used:
4ss′τa−s,−s′ = −τas′s , for s, s′ = ±1/2 . (F.30)
All the results of eqs. (F.23)–(F.26) can be derived in this manner.
F.3 Two-component Bouchiat-Michel formulae
To establish the Bouchiat-Michel formulae, we begin with the following identity:
12 (δss′ + ~σ ·sa τass′)
∑
t=±1/2
χtχ†t
= χs′χ†s. (F.31)
To verify eq. (F.31), we use eq. (F.20) to write ~σ ·saχt = τat′tχt′ and evaluated the product of
two Pauli matrices:
τass′τat′t = 2 δstδs′t′ − δss′δtt′ . (F.32)
Using eq. (F.21) and the completeness relation given in eq. (B.8), we can rewrite eq. (F.31) as:
χs′χ†s = 1
2
(δss′ +
1
m
√p·σ Sss′ ·σ
√p·σ), (F.33)
where Sss′ is defined in eq. (F.8). Hence, with both spinor indices in the lowered position,
x(~p, s′)x(~p, s) =√p·σ χs′χ†
s
√p·σ
= 12
√p·σ
[δss′ +
1
m
√p·σ Sss′ ·σ
√p·σ]√
p·σ
= 12
[p·σδss′ +
1
mp·σ Sss′ ·σ p·σ
]
= 12 [p·σδss′ −mSss′ ·σ] . (F.34)
In the final step of eq. (F.34), we simplified the product of three dot-products by noting that
p·Sa = 0 implies that Sa ·σ p·σ = −p·σ Sa ·σ. Eq. (F.34) is the two-component version of one of
the Bouchiat-Michel formulae. We list below a complete set of Bouchiat-Michel formulae, which
can be derived by similar techniques:
xα(~p, s′)xβ(~p, s) = 1
2(p δss′ −mSss′)·σαβ , (F.35)
yα(~p, s′)yβ(~p, s) = 12 (p δss′ +mSss′)·σαβ , (F.36)
xα(~p, s′)yβ(~p, s) = 1
2
(mδss′δα
β − [(σ ·Sss′) (σ ·p)]αβ), (F.37)
yα(~p, s′)xβ(~p, s) = 12
(mδss′δ
αβ + [(σ ·Sss′) (σ ·p)]αβ
). (F.38)
If we set s = s′, we recover eqs. (3.44)–(3.47) as expected.
152
An equivalent set of Bouchiat-Michel formulae can be obtained by raising and/or lowering
the appropriate free spinor indices using eqs. (2.22) and (2.61):
xα(~p, s′)xβ(~p, s) = 12(p δs′s −mSs′s)·σαβ , (F.39)
yα(~p, s′)yβ(~p, s) = 1
2(p δs′s +mSs′s)·σαβ , (F.40)
yα(~p, s′)xβ(~p, s) = − 1
2
(mδs′sδα
β + [(σ ·Ss′s) (σ ·p)]αβ), (F.41)
xα(~p, s′)yβ(~p, s) = − 12
(mδs′sδ
αβ − [(σ ·Ss′s) (σ ·p)]αβ
). (F.42)
These latter set of formulae can also be verified directly by using the explicit forms for the
two-component spinor wave functions. In this derivation, the spin labels in eqs. (F.39)–(F.42)
are reversed relative to those in eqs. (F.35)–(F.38) due to eq. (F.30).
Other combinations of spinor bilinears are possible. However, eqs. (F.16)–(F.19) imply that
the x and y spinors are related:
y(~p, s) = 2sx(~p,−s) , y(~p, s) = 2sx(~p,−s) . (F.43)
Using eq. (F.43), all possible spinor bilinears can be obtained from eqs. (F.35)–(F.42).
Note that eqs. (F.35)–(F.43) also apply to helicity spinor wave functions x(~p, λ) and y(~p, λ)
after replacing s, s′ with λ, λ′ and using the Sa as defined in eqs. (F.12)–(F.14). Strictly speaking,
all results involving the spinor wave functions obtained up to this point apply in the case of a
massive spin-1/2 fermion. If we take the massless limit, then the four-vector S 3 does not exist,
as its definition depends on the existence of a rest frame. (In contrast, the four-vectors S 1 and
S2 do exist in the massless limit.) Nevertheless, massless helicity spinor wave functions are
well-defined; explicit forms can be found in eqs. (3.35)–(3.38). Using these forms, one can derive
the Bouchiat-Michel formulae for a massless spin-1/2 fermion:
xα(~p, λ′)xβ(~p, λ) = ( 1
2 − λ) δλλ′ p·σαβ , (F.44)
yα(~p, λ′)yβ(~p, λ) = ( 12 + λ) δλλ′ p·σαβ , (F.45)
xα(~p, λ′)yβ(~p, λ) = − 1
2(12 − λ
′)(12 + λ) [(σ ·S12)(σ ·p)]α β , (F.46)
yα(~p, λ′)xβ(~p, λ) = 12 (1
2 + λ′)(12 − λ) [(σ ·S21)(σ ·p)]α β , (F.47)
where
S12 ≡ S12 ,−
12
= S1 − iS2 , S21 ≡ S−12 ,
12
= S1 + iS2 . (F.48)
An explicit representation is given by:
12σ ·S12 =
sinθ
2cos
θ
2e−iφ sin2 θ
2
−eiφ cos2 θ
2− sin
θ
2cos
θ
2
, 1
2σ ·S21 =
sinθ
2cos
θ
2−e−iφ cos2 θ
2
eiφ sin2 θ
2− sin
θ
2cos
θ
2
.
(F.49)
153
The equivalent set of Bouchiat-Michel formulae, obtained by raising and/or lowering the appro-
priate free spinor indices, is given by:
xα(~p, λ′)xβ(~p, λ) = ( 12 − λ) δλλ′ p·σαβ , (F.50)
yα(~p, λ′)yβ(~p, λ) = ( 1
2 + λ) δλλ′ p·σαβ , (F.51)
yα(~p, λ′)xβ(~p, λ) = − 1
2(12 + λ′)(1
2 − λ) [(σ ·S12)(σ ·p)]α β , (F.52)
xα(~p, λ′)yβ(~p, λ) = 12(1
2 − λ′)(12 + λ) [(σ ·S21)(σ ·p)]α β . (F.53)
As a check, one can verify that the above results follow from eqs. (F.35)–(F.42) by replacing
s with λ, setting mSaµ = pµ δa3, applying the mass-shell condition (p2 = m2), and taking the
m→ 0 limit at the end of the computation.
We now demonstrate how to use the Bouchiet-Michel formulae to evaluate helicity ampli-
tudes involving two equal-mass spin-1/2 fermions. A typical amplitude involving a fermion-
antifermion pair, evaluated in the center-of-mass frame of the pair has the generic structure:
z(~p, λ) Γ z′(−~p, λ′) , (F.54)
where z is one of the two-component spinor wave functions x, x y or y, and Γ is a 2× 2 matrix
(in spinor space) made up of products of the identity matrix, σ and σ. As an illustration, we
evaluate:
xα(~p, λ) Γαβ yβ(−~p, λ′) = 2λ′ Γαβ xβ(−~p,−λ′)xα(~p, λ) = 2λ′ Γαβσ0ββyβ(~p, λ)xα(~p, λ) , (F.55)
where we have used eqs. (B.31) and (F.43). We can now employ the Bouchiat-Michel formula
to convert the above result into a trace. By a similar computation, all expressions of the form
of eq. (F.54) can be expressed as a trace:
xα(~p, λ) Γαβ yβ(−~p, λ′) = λ′ Tr[Γσ0(mδλλ′ + σ ·Sλλ′ σ ·p)
], (F.56)
yα(~p, λ) Γαβ xβ(−~p, λ′) = −λ′ Tr
[Γσ0(mδλλ′ − σ ·Sλλ′ σ ·p)
], (F.57)
yα(~p, λ) Γαβ yβ(−~p, λ′) = λ′ Tr
[Γσ0(σ ·p δλλ′ +mσ ·Sλλ′)
], (F.58)
xα(~p, λ) Γαβ xβ(−~p, λ′) = −λ′ Tr
[Γσ0(σ ·p δλλ′ −mσ ·Sλλ′)
], (F.59)
after making use of eqs. (F.35) and (F.38). Similarly, there are four additional results that make
use of eqs. (F.39) and (F.42):
yα(~p, λ) Γαβ xβ(−~p, λ′) = λ′ Tr[Γσ0(mδλ′λ − σ ·Sλ′λ σ ·p)
], (F.60)
xα(~p, λ) Γαβ yβ(−~p, λ′) = −λ′ Tr
[Γσ0(mδλ′λ + σ ·Sλ′λ σ ·p)
], (F.61)
xα(~p, λ) Γαβ xβ(−~p, λ′) = −λ′ Tr
[Γσ0(σ ·p δλ′λ −mσ ·Sλ′λ)
], (F.62)
yα(~p, λ) Γαβ yβ(−~p, λ′) = λ′ Tr
[Γσ0(σ ·p δλ′λ +mσ ·Sλ′λ)
]. (F.63)
154
For amplitudes involving equal mass fermions (or equal mass antifermions), other combinations
of spinor bilinears appear in which one x-spinor above is replaced by a y-spinor or vice versa.
These amplitudes can be reduced to one of the eight listed above by using eq. (F.43).
The traces are easily evaluated using the results of Appendix A. Here, we apply the above
results to the amplitude for the decay Z0 → f f [see Section 6.2]. The corresponding center-of-
mass frame helicity amplitude is a linear combination of eqs. (F.56) and (F.57) with Γ = σ and
Γ = σ, respectively. Evaluating the corresponding terms, we find for Γ = σ,
x(~p, λ)σµy(−~p, λ′) = 2λ′[mgµ0δλλ′ + pµS0
λλ′ − p0Sµλλ′ − 2m(SµS0 − S0Sµ)λλ′], (F.64)
where we have used eq. (F.10) to replace the term with the Levi-Civita tensor. Similarly, we
calculate for Γ = σ,
y(~p, λ)σµx(−~p, λ′) = 2λ′[−mgµ0δλλ′ + pµS0
λλ′ − p0Sµλλ′ + 2m(SµS0 − S0Sµ)λλ′]. (F.65)
Eqs. (F.64) and (F.65) provide explicit forms for the Z 0 → f f decay helicity amplitudes defined
in eqs. (6.16) and (6.17).
The above method is not applicable if the two fermions have unequal mass. In order to
compute the helicity amplitudes of the form given by eq. (F.54) for unequal masses, a gener-
alization of the above techniques is required. Some methods for four-component spinor wave
functions have been proposed in ref. [137]. We leave it as an exercise for the reader to trans-
late these techniques so that they are applicable to helicity amplitudes expressed in terms of
two-component spinor wave functions.
F.4 Four-component Bouchiat-Michel formulae
Using the resuls of Appendix E, the translation of the results of the previous section
into four-component spinor notation is straightforward. First, we consider a massive spin-1/2
fermion. Eqs. (F.23)–(F.26) yield [141]:
γ5/Sa u(~p, s′) = τass′ u(~p, s) , γ5/Sa v(~p, s′) = τas′s v(~p, s) , (F.66)
u(~p, s′) γ5/Sa = τass′ u(~p, s) , v(~p, s′) γ5/Sa = τas′s v(~p, s) . (F.67)
In the case of a = 3, eqs. (F.66) and (F.67) reduce to those of eq. (E.19).
The four-component Bouchiat-Michel formulae [135–137] can be obtained from eqs. (F.35)–
(F.42):
u(~p, s′)u(~p, s) = 12 [δss′ + γ5/Sss′ ] (/p+m) , (F.68)
v(~p, s′)v(~p, s) = 12 [δs′s + γ5/Ss′s] (/p−m) , (F.69)
where Sss′ ≡ Saτass′. As expected, the above results for s = s′ correspond to the spin projection
operators given in eqs. (E.20) and (E.21). Related formulae involving products of u and v-spinors
155
can be obtained by using
v(~p, s) = −2sγ5u(~p,−s) , u(~p, s) = 2sγ5v(~p,−s) , (F.70)
which follow from eq. (F.43).
Eqs. (F.66)–(F.70) also apply to helicity u and v-spinors, after replacing s, s ′ with λ, λ′ and
using the Sa as defined in eq. (F.14). In the convention where the two-component spinor wave
function satisfies eq. (B.26), the four-component versions of eqs. (B.31)–(B.34) yield:
u(−p , −λ) = γ0 u(p , λ) , v(−p , −λ) = γ0 v(p , λ) , (F.71)
u(−p , −λ) = u(p , λ) γ0 , v(−p , −λ) = v(p , λ) γ0 . (F.72)
In order to consider the massless limit, one must employ helicity spinors, as discussed in
Appendix F.3. For a = 1, 2, eqs. (F.66) and (F.67) apply in the m → 0 limit as written. The
corresponding massless limit for the case of a = 3 is smooth and results in eq. (E.22). Similarly,
the massless limit of the Bouchiat-Michel formulae for helicity spinors can be obtained by setting
mSaµ = pµ δa3, applying the mass-shell condition (p2 = m2), and taking the m→ 0 limit at the
end of the computation. The end result is
u(p, λ′)u(p, λ) = 12 (1 + 2λγ5) /p δλλ′ + 1
2γ5[/S1τ1λλ′ + /S2τ2
λλ′ ] /p , (F.73)
v(p, λ′)v(p, λ) = 12 (1− 2λγ5) /p δλ′λ + 1
2γ5[/S1τ1λ′λ + /S2τ2
λ′λ] /p . (F.74)
As expected, when λ = λ′, we recover the helicity projection operators for massless spin-1/2
particles given in eqs. (E.23) and (E.24).
As before, we can use the Bouchiat-Michel formulae to evaluate helicity amplitudes involving
two equal-mass spin-1/2 fermions. A typical amplitude involving a fermion-antifermion pair,
evaluated in the center-of-mass frame of the pair, has the generic structure:
w(~p, λ) Γw′(−~p, λ′) , (F.75)
where w is either a u or v spinor, w′ is respectively either a v or u spinor, and Γ is a product of
Dirac gamma matrices. For example,
u(~p, λ) Γ v(−~p, λ′) = −2λ′u(~p, λ) Γ γ5 u(−~p,−λ′) = −2λ′u(~p, λ) Γ γ5 γ0 u(~p, λ′) , (F.76)
where we have used the results of eqs. (F.70) and (F.71). We can now employ the Bouchiat-
Michel formula to convert the above result into a trace. By a similar computation, all expressions
of the form of eq. (F.75) can be expressed as a trace:
u(~p, λ) Γ v(−~p, λ′) = −λ′ Tr[Γγ5γ
0(δλλ′ + γ5/Sλλ′)(/p+m)], (F.77)
v(~p, λ) Γu(−~p, λ′) = λ′ Tr[Γγ5γ
0(δλ′λ + γ5/Sλ′λ)(/p−m)]. (F.78)
156
These results are the four-component analogues of eqs. (F.56)–(F.59) and eqs. (F.60)–(F.63),
respectively. For amplitudes that involve a pair of equal mass fermions [or equal mass an-
tifermions], w and w′ in eq. (F.75) are both u-spinors [or v-spinors]. Using eq. (F.70), these
amplitudes can then be evaluated using the results of eqs. (F.77) and (F.78) above.
As an example, we consider once again the decay Z 0 → f f . The decay amplitude is equal
to eq. (F.77), where Γ is a linear combination of 12γ
µ(1 − γ5) and 12γ
µ(1 + γ5). Evaluating the
corresponding traces yields:
u(~p, λ) 12γ
µ(1− γ5) v(−~p, λ′) = 2λ′[mgµ0δλλ′ + pµS0
λλ′ − p0Sµλλ′ + iε0µνρ(Sλλ′)νpρ], (F.79)
u(~p, λ) 12γ
µ(1 + γ5) v(−~p, λ′) = 2λ′[−mgµ0δλλ′ + pµS0
λλ′ − p0Sµλλ′ − iε0µνρ(Sλλ′)νpρ]. (F.80)
Using eq. (F.10), we see that eqs. (F.79) and (F.80) reproduce exactly the results of eqs. (F.64)
and (F.65), respectively.
Finally, we note that if the two fermions do not have the same mass, then the method
presented above is not applicable. However, generalizations of the above method exist in the
literature that can be employed to evaluate helicity amplitudes of the form of eq. (F.75) for
unequal mass fermions; see, e.g., ref. [137].
Appendix G: The helicity amplitude techinique
In this appendix, we discuss how to apply our formalism to the helicity amplitude technique.
The latter is very useful when computing scattering cross sections for multi-particle final states,
which is typically done numerically. We shall review the formalism of Hagiwara and Zeppenfeld
(HZ) [63] and show how our formalism can be connected to theirs. In particular, we present a
translation between the two in Table 2.
After factoring the propagators an arbitrary tree amplitude with external fermions can be
expressed in terms of a “fermion string”
ψ1PR,L/a1/a2 . . . /anψ2 , (G.1)
where ψi denote the four -component spinor wave functions
ψi = u(pi, λi) , or v(pi, λi) , (G.2)
and PR,L = (1 ± γ5)/2 are the standard projection operators [cf. eq. (E.5)] in the chiral rep-
resentation. Furthermore, ai stands for an arbitrary Lorentz four vector, which can be a four-
momentum (pµi ), a vector boson wave-function (εµ(pi, λi)) an axial vector (εµνρκpνi pρjpκk) or another
fermion string with uncontracted Lorentz indices, e.g. ψ3γµψ4.
157
Our Formalism HZ Formalism
xα(p, λ) u(p, λ)−
xα(p, λ) −v∗(p, λ)+
xα(p, λ) −v(p, λ)+
xα(p, λ) u∗(p, λ)−
yα(p, λ) −v(p, λ)−
yα(p, λ) u∗(p, λ)+
yα(p, λ) u(p, λ)+
yα(p, λ) −v∗(p, λ)−
p · σ /p+
p · σ /p−
σµ σµ+
σµ σµ−
Table 2: Translation between our notation and Hagiwara and Zeppenfeld (HZ) [63].
In order to rewrite the fermion string, eq. (G.1), in terms of two-component spinors, we
need the HZ decomposition of the spinors (G.2)
ψi ≡
(ψi)−
(ψi)+
, u(pi, λi) ≡
u(pi, λi)−
u(pi, λi)+
, v(pi, λi) ≡
v(pi, λi)−
v(pi, λi)+
. (G.3)
The corresponding expressions in our notation are given in Table 2. Note the additional sign for
the v± spinors. This is because HZ take v(p, λ) = CuT (p, λ) with C = iγ2γ0 = −iγ0γ2, which
differs by a sign from our convention. In deriving the correspondence between the notations it
is helpful to note that
√p · σ χλ =
E +m− 2λ|~p|√2(E +m)
χλ = ω+(p)χλ , (G.4)
√p · σ χλ =
E +m+ 2λ|~p|√2(E +m)
χλ = ω−(p)χλ (G.5)
where we have used eq. (3.13) (labels) and, as in HZ, ω±(p) ≡√E ± |~p|. Note that we use our
notation: λ = ± 12 . In HZ λ = ±1.
After employing the Fierz identieties given by eqs. (2.55)–(2.57) to get rid of Lorentz in-
dices which are contracted between different fermion strings, the general fermion string can be
158
expressed as (in the notation of HZ)
FS = (ψ1)†α[a1, a2, . . . , an]
α(ψ2)δnα, (G.6)
with the 2-component spinor index α = ±, and δn = (−1)n+1. Furthermore
[a1, a2, . . . , an]α ≡ (/a1)α(/a2)−α . . . (/an)δnα (G.7)
where
(/a)± = aµσµ± . (G.8)
In our notation σµ+ = σµ and σµ− = σµ, (cf. Table 2). Using the formalism developed in
this paper and employing the Fierz identities given by eqs. (2.55)–(2.57), it is straightforward
to express any tree amplitude involving external two component fermions in terms of fermion
strings in the form (G.6). This is the immediate link between our work and HZ.
In order to see how to numerically compute amplitudes, we express FS in terms of the
relevant momenta. In the following, we use our sign convention for the spinors v±. We first
rewrite FS as [63]
FS = CiCjωα(2λi)(pi)ωα(2λj)(pj)S(pi, a1, a2, . . . , an, pj)αλiλj , (G.9)
where
Ck =
1 for (ψk)τ = u(pk, λk)τ ,
(2λk)τ for (ψk)τ = v(pk,−λk)τ ,τ = ±1, λ = ±1
2(G.10)
The function S can be expressed as
S(pi, a1, a2, . . . , an, pj)αλiλj =
[n∏
k=1
∑
τk=±(ak)−αδkτk
]T (pi, a1)(2λi)τ1T (a1, a2)τ1τ2
. . . T (an−1, an)τn−1τnT (an, pj)τn(2λj) (G.11)
where the functions T [58, 63] an be expressed entirely in terms of the relevant momenta.
T (a, b)++ = N−1ab
[(|~a|+ az)(|~b|+ bz) + (ax − iay)(bx + iby)
](G.12)
T (a, b)+− = N−1ab
[−(|~a|+ az)(bx − iby) + (ax − iay)(|~b|+ bz)
](G.13)
T (a, b)−+ = −T (a, b)∗+− (G.14)
T (a, b)−− = T (a, b)∗++ (G.15)
with the normalization factor given by
Nab =
√|~a|(|~a|+ az)|~b|(|~b|+ bz) . (G.16)
159
Two-component
fermion fields SU(3) SU(2)L Y T3 Q = T3 + 12Y
u
d
triplet
tripletdoublet
16
16
12
−12
23
−13
uc anti-triplet singlet − 23 0 −2
3
dc anti-triplet singlet 13 0 1
3
ν
e
singlet
singletdoublet
−12
−12
12
−12
0
−1
ec singlet singlet 1 0 1
Table 3: Fermions of the Standard Model and their SU(3)×SU(2)L×U(1)Y quantum numbers.
Appendix H: Standard Model fermion interaction vertices
In the Standard Model, one generation of quarks and leptons is described by the two-component
fermion fields listed in Table 3, where Y is the weak hypercharge, T3 is the third component of
the weak isospin, and Q = T3 + Y is the electric charge. After SU(2)L×U(1)Y breaking, the
quark and lepton fields gain mass in such a way that the above two-component fields combine
to make up four-component Dirac fermions:
U =
u
uc
, D =
d
dc
, E =
e
ec
, (H.1)
while the neutrino remains massless. (The extension of the Standard Model to include neutrino
mass will be treated elsewhere.)
Here, we follow the convention for particle symbols established in Table 1. Note that u and
d are two-component fields, whereas the usual four-component quark and charged lepton fields
are denoted by capital letters U , D and E. Consider a generic four-component field expressed
in terms of the corresponding two-component fields:
F =
f
f c
. (H.2)
160
ga
µ
qk
qj
−igsT akj σαβµ
α
β
ga
µ
(qc)j
(qc)k
igsTakj σαβµ
α
β
Figure 66: Fermionic Feynman rules for QCD that involve the gluon, with q = u, d, c, s, t, b.Lowered (raised) indices j, k correspond to the fundamental (anti-fundamental) representation
of SU(3)c. For each rule, a corresponding one with lowered spinor indices is obtained by σ αβµ →−σµβα.
The electroweak quantum numbers of f are denoted by T f3 , Yf and Qf , whereas the corre-
sponding quantum numbers for f c are T fc
3 = 0 and Qfc = Yfc = −Qf . Thus we have the
correspondence to our general notation [eq. (E.4)]
f ←→ χ, f c ←→ η . (H.3)
We can then immediately translate the couplings given in the general case in Fig. 9 to the
Standard Model.
The QCD color interactions of the quarks are governed by the following interaction La-
grangian:
Lint = −gsAµa q ji σµ(T a)jkqki , (H.4)
summed over the generations i, where q is a (mass-eigenstate) quark field, j and k are SU(3) color
labels and T a are the color generators in the triplet representation of SU(3). The corresponding
Feynman rules are given in Fig. 66.
Next, we write out the Feynman rules for the electroweak interactions of quarks and leptons.
Consider the charged current interactions of the quarks:
Lint = − g√2
[¯uiσµdiW
+µ + d iσµuiW
−µ
], (H.5)
where the hatted symbols indicate interaction eigenstates and i labels the generations. Following
the discussion of Appendix E, we convert to mass eigenstates for the quarks. That is, we
introduce four unitary matrices, Lu, Ld, Ru and Rd, [cf. eq. (3.80)] such that
ui = (Lu)ijuj , di = (Ld)i
jdj , uci = (Ru)ijucj , and dci = (Rd)
ijdcj , (H.6)
where the unhatted fields u, d, uc and dc are the corresponding mass eigenstates. It then follows
that ¯uiσµdi = Kijuiσµdj , where
K = L†uLd (H.7)
161
γf
f
−ieQfσαβµ
α
β
µ
γf c
f c
ieQfσαβµ
α
β
µ
Z f
f
−i gcW
(T f3 − s2WQf )σαβµ
α
β
µ
Z f c
f c
−igs2W
cWQfσ
αβµ
α
β
µ
W− uj
di
− i√2g[K†]ijσ
αβµ
α
β
µ
W+dj
ui
− i√2g[K ]i
jσαβµ
α
β
µ
W−ν`
`
− i√2gσαβµ
α
β
µ
W+`
ν`
− i√2gσαβµ
α
β
µ
Figure 67: Feynman rules for the two-component fermion interactions with electroweak gaugebosons in the Standard Model. For the W bosons, the charge indicated is flowing into thevertex. The electric charge is denoted by Qf , with Qe = −1 for the electron, and T f3 is 1/2for up-type quarks and neutrinos and is −1/2 for down-type quarks and charged leptons. TheCKM mixing matrix is denoted K, and sW ≡ sin θW , cW ≡ cos θW and e ≡ g sin θW . For eachrule, a corresponding one with lowered spinor indices is obtained by σ αβµ → −σµβα.
is the unitary Cabibbo-Kobayashi-Maskawa (CKM) matrix.64 The charged current interactions
take the form
Lint = − g√2
[Ki
j uiσµdjW+µ + (K†)i
j d iσµujW−µ + νiσµeiW
+µ + eiσµνiW
−µ
], (H.8)
where
(K†)ij ≡Kj
i ≡ (Kji)∗. (H.9)
We have also included the leptons, which do not mix. (Note that the Standard Model does not
have W± interactions with uc and dc.) The corresponding Feynman rules are given in Fig. 67.
64The CKM matrix elements Vij as defined in ref. [144] are related by, for example, Vtb = K33 and Vus = K1
2.
162
hSM
f cj
fi
−i√2Yfiδ
ij δα
β
α
β
hSM
f cj
fi
−i√2Yfiδ
ji δ
αβ
α
β
Figure 68: Feynman rules for the Standard Model Higgs boson interactions with fermions.
The corresponding interaction of the fermions with the neutral gauge bosons are also given
in Fig. 67. The neutral current interactions are flavor-conserving. For each of the rules of Fig. 67,
we have chosen to employ σαβµ . If the indices are lowered one should take σαβµ → −σµβα.
The Yukawa interactions of the fermions with the Higgs field are given by:
−LY = (Y u)ij
[Φ0uiu
cj − Φ+diucj]
+ (Y d)ij
[Φ−uid
cj + Φ0∗didcj]
+ c.c. (H.10)
The Higgs fields can be written in terms of the physical Higgs scalar hSM and Nambu-Goldstone
bosons G0, G± as
Φ0 = v +1√2(hSM + iG0) (H.11)
Φ+ = G+ = (Φ−)∗ = (G−)∗. (H.12)
where v =√
2mW/g ≈ 175 GeV. In the unitary gauge appropriate for tree-level calculations,
the Nambu-Goldstone bosons become infinitely heavy and decouple. After diagonalization of
the quark mass matrices,
(Mu)ij = v(Y u)
ij , (M d)
ij = v(Y d)
ij , (H.13)
one obtains LTuMuRu = diag(mu,mc,mt) and LT
dMdRd = diag(md,ms,mb). The resulting
Higgs-fermion interactions are diagonal as shown in Fig. 68. Here, the diagonalized Higgs-
fermion Yukawa coupling matrices appear:
diag(Yu1, Yu2, Yu3) ≡ diag(Yu, Yc, Yt) = LT
uY uRu (H.14)
diag(Yd1, Yd2, Yd3) ≡ diag(Yd, Ys, Yb) = LT
dY dRd. (H.15)
Likewise, we define
Ye1 ≡ Ye, Ye2 ≡ Yµ, Ye3 ≡ Yτ (H.16)
for the (unmixed) leptons. Note that bold-faced symbols are used for the non-diagonal Yukawa
matrices, while non-bold-faced symbols are used for the diagonalized Yukawa couplings. The
latter are related to the corresponding fermion masses by Yfi = mfi/v, where i labels the fermion
generation. The corresponding interaction Lagrangian is
Lint =1√2hSM
[Yuiuiu
ci + Ydididci + Yeieie
ci]+ c.c. (H.17)
163
G0
ucj
ui
Yui√2δij δα
β
α
βG0
ucj
ui
−Yui√2δji δ
αβ
α
β
G0
dcj
di
−Ydi√2δij δα
β
α
βG0
dcj
di
Ydi√2δji δ
αβ
α
β
G0
`c
`
− Y`√2δαβ
α
βG0
`c
`
Y`√2δαβ
α
β
G+
uci
dj
iYui[K]ij δα
β
α
βG+
uci
dj
iYui[K†]j
i δαβ
α
β
G−
dci
uj
−iYdi[K†]ij δα
β
α
βG+
dci
uj
−iYdi[K]ji δαβ
α
β
G−
`c
ν`
−iY` δαβ
α
βG+
`c
ν`
−iY` δαβ
α
β
Figure 69: Feynman rules for the Standard Model Nambu-Goldstone boson interactions withquarks and leptons.
164
In the case of more general covariant gauge-fixing, including Feynman gauge or Landau
gauge, the Goldstone bosons appear explicitly in internal lines of Feynman diagrams. The
Feynman rules for G0-fermion interactions are flavor diagonal, whereas the corresponding rules
for G± have a flavor-changing component that depends on the CKM matrix elements. The
relevant interaction Lagrangian for quarks follows from eqs. (H.10) and (H.15):
Lint =i√2
[Ydidid
ci − Yuiuiuci]G0 + Yui[K]i
jdjuciG+ − Ydi[K†]i
jujdciG− + c.c. (H.18)
The resulting diagrammatic Feynman rules are shown in Fig. 69, together with the corresponding
ones for the leptons.
165
Appendix I: MSSM Fermion Interaction Vertices
I.1 Higgs-fermion interaction vertices in the MSSM
The MSSM Higgs sector is a two-Higgs-doublet model containing eight real scalar degrees of
freedom: one complex Y = −1/2 doublet, Hd = (H0d ,H
−d ) and one complex Y = +1/2 doublet,
Hu = (H+u ,H
0u). The notation reflects the form of the MSSM Higgs sector coupling to fermions:
H0d [H0
u] couples exclusively to down-type [up-type] fermion pairs. In the supersymmetric model,
both hypercharge Y = −1/2 and Y = +1/2 complex Higgs doublets are required in order that
the theory (which now contains the corresponding higgsino superpartners) remain anomaly-free.
The supersymmetric structure of the theory also requires (at least) these two Higgs doublets to
generate mass for both “up”-type and “down”-type quarks and charged leptons.
To find the couplings of the Higgs fields, we expand them around vacuum expectation values
vd and vu. Depending on the application, these may be chosen to be the minimum of the tree-
level potential, or of the full loop-corrected effective potential, or just left arbitrary. The phases
of the Higgs fields are chosen such that vu and vd are real and positive. That is, the tree-level
MSSM Higgs sector conserves CP, which implies that the neutral Higgs mass eigenstates have
definite CP quantum numbers. Spontaneous electroweak symmetry breaking results in three CP-
odd Goldstone bosons G±, G0, which are absorbed and become the longitudinal components
of the W± and Z. The remaining five physical Higgs particles consist of a charged Higgs pair
H±, one CP-odd scalar A0, and and two CP-even scalars h0 and H0. One can parameterize the
mixing angles between Higgs gauge eigenstates and mass eigenstates by writing:
H0u = vu +
1√2
∑
φ0
kuφ0φ0, (I.1)
H0d = vd +
1√2
∑
φ0
kdφ0φ0, (I.2)
H+u =
∑
φ+
kuφ+φ+, (I.3)
H−∗d =
∑
φ+
kdφ+φ+, (I.4)
where, for φ0 = (h0, H0, G0, A0),
kuφ0 = (cosα, sinα, i sinβ0, i cos β0) (I.5)
kdφ0 = (− sinα, cosα, −i cos β0, i sinβ0) (I.6)
and for φ+ = (G+, H+),
kuφ+ = (sinβ±, cosβ±), (I.7)
kdφ+ = (− cos β±, sinβ±). (I.8)
166
Here the normalization is such that, if one chooses vu, vd to be near the true minimum of the
Higgs effective potential, then v2 ≡ v2d + v2
u = 2m2W /g
2 ≈ (175 GeV)2. Note that in the special
case that vu and vd are the minimum of the tree-level potential, the mixing angles β± in the
charged Higgs sector and β0 in the pseudo-scalar Higgs sectors coincide exactly with
β ≡ arctan(vu/vd). (I.9)
However, if one expands around a more general choice of vu, vd, including for example the
minimum of the full effective potential, then the tree-level mixing angles β0 and β± are distinct
from each other and from β. (Depending on the choice of renormalization scale for a particular
calculation, the tree-level potential in the MSSM may have a very different minimum from
the true minimum of the full effective potential, or may not have a proper minimum at all.)
Therefore, we do not assume anything specific about vu and vd except that they are real and
positive.
The Higgs-quark Yukawa couplings in the gauge-interaction basis are given by:
−L = (Y u)ij
[uiu
cjH0u − diucjH+
u
]+ (Y d)
ij
[did
cjH0d − uidcjH−
d
]+ c.c. (I.10)
Let us change to the mass-eigenstate basis by using eq. (H.6) and (I.1)-(I.4). After diagonal-
ization of the fermion mass matrices, (M u)ij = vu(Y u)
ij and (Md)
ij = vd(Y d)
ij one ob-
tains LTuMuRu = diag(mu,mc,mt) and LT
dMdRd = diag(md,ms,mb). The resulting neutral
Higgs-fermion interactions are diagonal. Here, the diagonalized Higgs-fermion Yukawa coupling
matrices appear:65
diag(Yu1, Yu2, Yu3) ≡ diag(Yu, Yc, Yt) = LT
uY uRu , (I.11)
diag(Yd1, Yd2, Yd3) ≡ diag(Yd, Ys, Yb) = LT
dY dRd , (I.12)
and, for the leptons,
Ye1 = Ye, Ye2 = Yµ, Ye3 = Yτ . (I.13)
The diagonalized Yukawa couplings are related to the corresponding fermion masses by
Ydi = mdi/vd , Yei = mei/vd , Yui = mui/vu . (I.14)
The interactions of the neutral Higgs scalars φ0 = (h0,H0, G0, A0) with Standard Model
fermions are given in Fig. 70. Note that the last two rules involve kdφ0 and kuφ0 , while the first
65We have used the same symbol for the Yukawa couplings in the MSSM as we did for the Standard ModelYukawa couplings in Appendix H. However, it is important to note that they are normalized differently becauseof the presence of two Higgs VEVs. If we use a superscript SM to distinguish the Standard Model Yukawacouplings of Appendix H, then the MSSM Yukawa couplings defined here are related by Yui = Y SM
ui / sin β andYdi = Y SM
di / cos β and Yei = Y SMei / cos β.
167
φ0
ucj
ui
− i√2Yuikuφ0δij δα
β
α
β
φ0
ucj
ui
− i√2Yuik
∗uφ0δ
ji δ
αβ
α
β
φ0
dcj
di
− i√2Ydikdφ0δij δα
β
α
β
φ0
dcj
di
− i√2Ydik
∗dφ0δ
ji δ
αβ
α
β
φ0
`c
`
− i√2Y`kdφ0 δα
β
α
β
φ0
`c
`
− i√2Y`k
∗dφ0 δ
αβ
α
β
Figure 70: Feynman rules for the interactions of neutral Higgs bosons φ0 = (h0,H0, G0, A0)with fermion-antifermion pairs in the MSSM. The repeated index j is not summed.
two rules involve their complex conjugates. This means that for h0 and H0, starting with the
rule with undotted fermion indices, one obtains the corresponding rule with dotted indices (with
the direction of the arrows reversed) by taking δαβ → δαβ. The situation for the pseudoscalar A0
(and G0) is different because kuA0 and kdA0 (and kuG0 and kdG0) are purely imaginary. Starting
with the rules for pseudoscalar interactions with fermions with undotted fermion indices, one
obtains the corresponding rule with dotted indices (with the direction of the arrows reversed)
by taking δαβ → −δαβ. The minus sign in the last operation is a signal that A0 and G0 are
CP-odd scalars.66
The couplings of the charged Higgs boson to quark-antiquark pairs are not flavor diagonal
and involve the CKM matrix K. Starting with eq. (I.10), and changing to the mass-eigenstate
basis as before, one obtains
Lint = (LT
dY uRu)ijdiu
cjH+ cosβ + (LT
uY dRd)ijuid
cjH− sinβ + h.c., (I.15)
and the corresponding G± interactions by taking sinβ → − cos β and cosβ → sinβ. Using
eqs. (H.7) and (I.12), one obtains (LT
dY uRu)ij = [K]j
iYuj and (LTuY dRd)
ij = [K†]j iYdj , with
no sum on repeated indices. The resulting charged-scalar Feynman rules are given in Fig. 71.
66Because the Feynman rules for A0 and G0 arise from a term in Lint proportional to i Im H0, the latter i flipssign when the rule is conjugated resulting in the extra minus sign noted above. As an additional consequence,noting that the Feynman rules are obtained from iLint, the overall A0 and G0 rules are real.
168
φ+
di
ucj
iYuj[K]jikuφ+ δα
β
α
β
φ−di
ucj
iYuj[K†]ijkuφ+ δαβ
α
β
φ−
dcj
ui
iYdj [K†]jikdφ+ δα
β
α
β
φ+
dcj
ui
iYdj [K]ijkdφ+ δαβ
α
β
φ−
`c
ν`
iY`kdφ+ δαβ
α
β
φ+
`c
ν`
iY`kdφ+ δαβ
α
β
Figure 71: Feynman rules for the interactions of charged Higgs bosons φ± = (G±,H±) withfermion-antifermion pairs in the MSSM.
169
I.2 Gauge interaction vertices for neutralinos and charginos
Following eqs. (C83) and (C88) of ref. [48], we define:
OLij = − 1√2Ni4V
∗j2 +Ni2V
∗j1 , (I.16)
ORij = 1√2N∗i3Uj2 +N∗
i2Uj1 , (I.17)
O′Lij = −Vi1V ∗
j1 − 12Vi2V
∗j2 + δijs
2W , (I.18)
O′Rij = −U∗
i1Uj1 − 12U
∗i2Uj2 + δijs
2W , (I.19)
O′′Lij = −O′′R∗
ij = 12 (Ni4N
∗j4 −Ni3N
∗j3) . (I.20)
Here U and V are the unitary matrices that diagonalize the chargino mass matrix:
U∗Mψ±V −1 = diag(m eC1,m eC2
) , (I.21)
with
Mψ± =
M2 gvu
gvd µ
. (I.22)
Also, N is a unitary matrix that diagonalizes the neutralino mass matrix,
N∗Mψ0N−1 = diag(m eN1,m eN2
,m eN3,m eN4
) , (I.23)
with
Mψ0 =
M1 0 −g′vd/√
2 g′vu/√
2
0 M2 gvd/√
2 −gvu/√
2
−g′vd/√
2 gvd/√
2 0 −µ
g′vu/√
2 −gvu/√
2 −µ 0
. (I.24)
We now list the gauge boson interactions with the neutralinos and charginos. The Feynman
rules for Z and γ interactions with charginos and neutralinos are given in Fig. 72 and the
corresponding rules for W± interactions are given in Fig. 73. For each of these rules, one
has a version with lowered spinor indices by replacing σ αβµ → −σµβα. We label fermion lines
with the symbols of the two-component fermion fields as given in Table 1. Note that the
ZNiNj interaction vertex also subsumes the O ′′Rij interaction found in four-component Majorana
Feynman rules as in ref. [48], due to the result of eq. (E.50) and the relation O ′′Rij = −O′′L
ji of
eq. (I.20).
170
γχ+j
χ+i
−ie δijσαβµ
α
β
µ
γχ−j
χ−i
ie δijσαβµ
α
β
µ
Z χ+j
χ+i
ig
cWO
′Lij σ
αβµ
α
β
µ
Z χ−i
χ−j
−i gcW
O′Rij σ
αβµ
α
β
µ
Z χ0j
χ0i
ig
cWO
′′Lij σ
αβµ
α
β
µ
Figure 72: Feynman rules for the chargino and neutralino interactions with neutral gaugebosons. The coupling matrices are defined in eqs. (I.18)-(I.20). For each rule, a corresponding
one with lowered spinor indices is obtained by σαβµ → −σµβα.
W−χ+j
χ0i
ig OLijσ
αβµ
α
β
µ
W−χ0i
χ−j
−ig ORijσ
αβµ
α
β
µ
W+χ0i
χ+j
ig OL∗ij σ
αβµ
α
β
µ
W+χ−j
χ0i
−ig OR∗ij σ
αβµ
α
β
µ
Figure 73: Feynman rules for the chargino and neutralino interactions with W ± gauge bosons.The charge indicated on the W boson is flowing into the vertex in each case. The couplingmatrices are defined in eqs. (I.16) and (I.17). For each rule, a corresponding one with lowered
spinor indices is obtained by σαβµ → −σµβα.
171
φ0χ0j
χ0i
−iY φ0χ0iχ
0j δα
β
α
β
φ0χ+j
χ−i
−iY φ0χ−i χ
+j δα
β
α
β
φ− χ+j
χ0i
−iY φ−χ0iχ
+j δα
β
α
β
φ+χ−j
χ0i
−iY φ+χ0iχ
−j δα
β
α
β
Figure 74: Feynman rules for the interactions of Higgs bosons φ0 = (h0,H0, G0, A0) andφ± = (G±,H±) with chargino-neutralino pairs. For each rule, there is a corresponding one withall arrows reversed, undotted indices changed to dotted indices with the opposite height, andthe Y coupling (without the explicit i) replaced by its complex conjugate.
I.3 Higgs interactions with charginos and neutralinos
The couplings of chargino and neutralino mass eigenstates to the Higgs mass eigenstates can be
written, in terms of the Higgs mixing parameters of eqs. (I.5) and (I.6) and the neutralino and
chargino mixing matrices of the previous subsection, as:
Y φ0χ0iχ
0j =
1
2(k∗dφ0N
∗i3 − k∗uφ0N
∗i4)(gN
∗j2 − g′N∗
j1) + (i↔ j), (I.25)
Y φ0χ+i χ
−j =
g√2(k∗uφ0V
∗i2U
∗j1 + k∗dφ0V
∗i1U
∗j2) (I.26)
Y φ+χ0iχ
−j = kdφ+
[g(N∗
i3U∗j1 −
1√2N∗i2U
∗j2)−
g′√2N∗i1U
∗j2
](I.27)
Y φ−χ0iχ
+j = kuφ+
[g(N∗
i4V∗j1 +
1√2N∗i2V
∗j2) +
g′√2N∗i1V
∗j2
]. (I.28)
We list the Higgs boson interactions with the neutralinos and charginos in Fig. 74. For each
of the Feynman rules in Fig. 74, one can reverse all arrows by taking δαβ → δαβ and complex
conjugating the corresponding coupling (but not the overall factor of i).
172
I.4 Chargino and neutralino interactions with fermions and sfermions
In the MSSM, scalar partners of the two-component fields q and qc are the squarks, denoted by
qL and qR, respectively. In our notation, q∗L and q∗R denote both the complex conjugate fields
and the names of the corresponding anti-squarks. Thus u, uL and uR all have electric charges
+2/3, whereas uc, u∗L and u∗R all have electric charges −2/3. Likewise, the scalar partners of
the two-component fields ` and `c are the charged sleptons, denoted by ˜L and ˜R, respectively,
with ` = e, µ, τ . The sneutrino, ν is the superpartner of the neutrino. There is no νR, since
there is no νc in the theory.
The Feynman rules for the chargino-quark-squark interactions are given in Fig. 75, and
the rules for the neutralino-quark-squark interactions are given in Fig. 76. (Note that chargino
interaction vertices involving ucdR and dcuR do not occur in the MSSM.) Here we have taken
the quark and lepton two-component fields to be in a mass-eigenstate basis, and the squark and
slepton field basis consists of the superpartners of these fields, as described above. Therefore,
in practical applications, one must include unitary rotation matrix elements relating the the
squarks and sleptons as given to the mass eigenstates, which can be different.
In principle, all sfermions with a given electric charge can mix with each other. However,
there is a popular, and perhaps phenomenologically and theoretically favored, approximation in
which only the sfermions of the third family have significant mixing. For f = t, b, τ , one can
then write the relationship between the gauge eigenstates fL, fR and the mass eigenstates f1,
f2 as
fRfL
= Xf
f1
f2
(I.29)
where
Xf =
Rf1 Rf2Lf1 Lf2
(I.30)
is a 2× 2 unitary matrix. Then one can choose Rf1= L∗
f2= cf , and Lf1 = −R∗
f2= sf with
|cf |2 + |sf |2 = 1. (I.31)
If there is no CP violation, then cf and sf can be taken real, and they are the cosine and sine
of a sfermion mixing angle.67 For the other charged sfermions (f = u, d, c, s, e, µ), one can use
the same notation, and approximate Lf2 = Rf1 = 1 and Lf1 = Rf2 = 0. The resulting Feynman
rules for squarks and sleptons that mix within each generation are shown in Figs. 77 and 78.
67Our convention for the sfermion mixing has the property that for zero mixing angle, ef1 = efR and ef2 = efL.The conventions most commonly found in the literature unfortunately do not have this nice property.
173
dLj uk
χ−i
−igU∗i1[K
†]kj δα
β
α
β
uLj dk
χ+i
−igV ∗i1[K]j
k δαβ
α
β
˜L ν`
χ−i
−igU∗i1 δα
β
α
β
ν` `
χ+i
−igV ∗i1 δα
β
α
β
dLj uck
χ+i
iV ∗i2[K]k
jYuk δαβ
α
β
uLjdck
χ−i
iU∗i2[K
†]jkYdj δαβ
α
β
dRj uk
χ−i
iU∗i2[K
†]kjYdj δα
β
α
β
uRj dk
χ+i
iV ∗i2[K]j
kYuj δαβ
α
β
˜R ν`
χ−i
iU∗i2Y` δα
β
α
β
ν` `c
χ−i
iU∗i2Y` δα
β
α
β
Figure 75: Feynman rules for the interactions of charginos with fermion/sfermion pairs inthe MSSM. The fermions are taken to be in a mass-eigenstate basis, and the sfermions are ina basis whose elements are the supersymmetric partners of them. For each rule, there is acorresponding one with all arrows reversed, undotted indices changed to dotted indices with theopposite height, and the coupling (without the explicit i) replaced by its complex conjugate. Analternative version of these rules, for the case that mixing is allowed only among third-familysfermions, is given in Fig. 77.
174
fLj fk
χ0i
−i√
2[gT f3 N
∗i2 + g′(Qf − T f3 )N∗
i1
]δkj δα
β
α
β
fRj f ck
χ0i
i√
2g′QfN∗i1 δ
jkδα
β
α
β
uLjuck
χ0i
−iN∗i4Yujδ
jk δα
β
α
β
uRj uk
χ0i
−iN∗i4Yujδ
kj δα
β
α
β
dLj dck
χ0i
−iN∗i3Ydjδ
jk δα
β
α
β
dRj dk
χ0i
−iN∗i3Ydjδ
kj δα
β
α
β
˜L `c
χ0i
−iN∗i3Y` δα
β
α
β
˜R `
χ0i
−iN∗i3Y` δα
β
α
β
Figure 76: Feynman rules for the interactions of neutralinos with fermion/sfermion pairs inthe MSSM. The fermions are taken to be in a mass-eigenstate basis, and the sfermions are ina basis whose elements are the supersymmetric partners of them. For each rule, there is acorresponding one with all arrows reversed, undotted indices changed to dotted indices with theopposite height, and the coupling (without the explicit i) replaced by its complex conjugate. Analternative version of these rules, for the case that mixing is allowed only among third-familysfermions, is given in Fig. 78.
175
tj b
χ+i
i(YtV∗i2R
∗tj− gV ∗
i1L∗tj
) δαβ
α
β
tj bc
χ−i
iYbU∗i2Ltj δα
β
α
β
bj t
χ−i
i(YbU∗i2R
∗bj− gU∗
i1L∗bj
) δαβ
α
β
bj tc
χ+i
iYtV∗i2Lbj δα
β
α
β
ντ τ
χ+i
−igV ∗i1 δα
β
α
β
ντ τ c
χ−i
iYτU∗i2 δα
β
α
β
τj ντ
χ−i
i(YτU∗i2R
∗τj− gU∗
i1L∗τj
) δαβ
α
β
Figure 77: Feynman rules for the interactions of charginos with third-family fermion/sfermionpairs in the MSSM. The fermions are taken to be in a mass-eigenstate basis. CKM mixing isneglected, and the sfermions are assumed to only mix within the third family. The correspondingrules for the first and second families with the approximation of no mixing and vanishing fermionmasses can be obtained from these by setting Yf = 0 and Lf2 = Rf1 = 1 and Lf1 = Rf2 = 0 (so
that f1 = fR and f2 = fL). For each rule, there is a corresponding one with all arrows reversed,undotted indices changed to dotted indices with the opposite height, and the coupling (withoutthe explicit i) replaced by its complex conjugate.
176
tj t
χ0i
−i[YtN
∗i4R
∗tj
+ 1√2(gN∗
i2 + g′N∗i1/3)L
∗tj
]δαβ
α
β
tj tc
χ0i
−i[YtN
∗i4Ltj −
2√
23 g′N∗
i1Rtj
]δαβ
α
β
bj b
χ0i
−i[YbN
∗i3R
∗bj
+ 1√2(−gN∗
i2 + g′N∗i1/3)L
∗bj
]δαβ
α
β
bj bc
χ0i
−i[YbN
∗i3Lbj +
√2
3 g′N∗
i1Rbj
]δαβ
α
β
ντ ντ
χ0i
− i√2(gN∗
i2 − g′N∗i1) δα
β
α
β
τj τ
χ0i
−i[YτN
∗i3R
∗τj− 1√
2(gN∗
i2 + g′N∗i1)L
∗τj
]δαβ
α
β
τj τ c
χ0i
−i[YτN
∗i3Lτj +
√2g′N∗
i1Rτj]δαβ
α
β
Figure 78: Feynman rules for the interactions of neutralinos with third-family fermion/sfermionpairs in the MSSM. The same comments apply as for Fig. 77.
177
For each of the Feynman rules in figs. 75-78, one can reverse all arrows by taking δαβ → δαβ
and taking the complex conjugate of the corresponding rule (but leaving the explicit factor of i
intact).
178
ga
µ
qk
qj
−igsT akj σαβµ
α
β
ga
µ
(qc)j
(qc)k
igsTakj σαβµ
α
β
gd
µ
gb
ga
−gsfabd σαβµ
α
β
Figure 79: Fermionic Feynman rules for SUSY QCD that involve the gluon, with q =u, d, c, s, t, b. Lowered (raised) indices j, k correspond to the fundamental (anti-fundamental)representation of SU(3)c. For each rule, a corresponding one with lowered spinor indices is
obtained by σαβµ → −σµβα.
I.5 SUSY QCD Feynman Rules
In two component notation, the Lagrangian governing the gluon’s interactions with fermions,
which come from the covariant derivatives in the kinetic terms, is
L = igsfabd (ga σµ gb)A
µd − gsT akj
∑
q
[qjσµqk − (qc)kσµ(q
c)j]Aµa . (I.32)
Here gs is the strong coupling constant, a, b, d = 1, 2, . . . , 8 are SU(3)c adjoint representation
indices, and fabd are the SU(3) structure constants. Raised (lowered) indices j, k = 1, 2, 3 are
color indices in the fundamental (anti-fundamental) representation. We have denoted the 2-
component gluino field by ga as in Table 1 and the gluon field by Aµa . The sum∑
q is over the
six flavors q = u, d, s, c, b, t (in either the mass-eigenstate or electroweak gauge-eigenstate basis).
The corresponding Feynman rules are shown in Fig. 79. The gluino-squark-quark Lagrangian is
L = −√
2gsTakj
∑
q
[gaqk q
∗jL + gaq
j qLk − ga(qc)j qRk − ga(qc)k q∗jR], (I.33)
where the squark fields are taken to be in the same basis as the quarks. The Feynman rules
resulting from these Lagrangian terms are shown in Fig. 80.
For practical applications, one typically takes the quark fields as the familiar mass eigen-
states, and then does a unitary rotation on the squarks in the corresponding basis to obtain
their mass eigenstate basis. In the approximation described above, in the paragraph containing
eqs. (I.29)-(I.31), one obtains the Feynman rules of Fig. 81, as an alternative to those of Fig. 80.
179
qLj ga
qk
−i√
2gsTakj δα
β
α
β
qLk ga
qj
−i√
2gsTakj δαβ
α
β
q∗kR ga
(qc)j
i√
2gsTakj δα
β
α
β
q∗jR ga
(qc)k
i√
2gsTakj δαβ
α
β
Figure 80: Fermionic Feynman rules for SUSY QCD that involve the squarks, in a basis corre-sponding to the quark mass eigenstates q = u, d, c, s, t, b. Lowered (raised) indices j, k correspondto the fundamental (anti-fundamental) representation of SU(3)c, and the index a labels the ad-joint representation carried by the gluino. The spinor index heights can be exchanged in eachcase, by replacing δα
β → δβα or δαβ → δβ α. For an alternative set of rules, incorporating LR
mixing, see Fig. 81.
qxj ga
qk
−i√
2gsL∗qxT akj δα
β
α
β
qxk ga
qj
−i√
2gsLqxTakj δαβ
α
β
q∗kx ga
(qc)j
i√
2gsRqxTakj δα
β
α
βq∗jx ga
(qc)k
i√
2gsR∗qxT akj δαβ
α
β
Figure 81: Fermionic Feynman rules for SUSY QCD that involve the squarks in the masseigenstate basis labeled by x = 1, 2 and q = u, d, c, s, t, b, in the approximation that mixing isallowed only within a given flavor (typically, for the third family only), as in eq. (I.29). Lowered(raised) indices j, k correspond to the fundamental (anti-fundamental) representation of SU(3)c,and the index a labels the adjoint representation carried by the gluino. The spinor index heightscan be exchanged in each case, by replacing δα
β → δβα or δαβ → δβ α.
180
Appendix J: Trilinear R-Parity Violating Fermion InteractionVertices
In the case of R-parity violation [145], the MSSM superpotential is extended by the following
terms [146]:
W 6Rp = 12λijkεabL
aiL
bjEk + λ′ijkεabL
aiQ
bjDk + 1
2λ′′ijkεc1c2c3U
c1i D
c2j D
c3k + κiεabL
aiH
bu . (J.1)
Here λijk, λ′ijk, λ
′′ijk are dimensionless coupling constants and i, j, k are generation indices. a, b =
1, 2 and c1, c2, c3 = 1, 2, 3 are SU(2) and SU(3) indices, respectively. Li, Qi are the lepton and
quark SU(2)-doublet left-chiral superfields. E i, U i, Di are the charged lepton and quark SU(2)-
singlet left-chiral superfields. κi is a mass-dimension one parameter, which leads to mixing
between the sleptons and Higgs fields, as well as between the leptons and Higgsinos. This
modifies the Feynman rules of Appendix I through additional mixing matrices, which we do not
include here [99]. Recently, the two-component fermion Feynman rules for the neutral fermions
have been given in [147]. The intractions in eq. (J.1) can significantly alter the phenomenology
at colliders (see for example [102,105]), in particular, since the LSP is no longer stable. See the
computations in Sects. 6.19, and 6.20.
The tri-linear terms in eq. (J.1) lead to additional Yukawa couplings as follows:
LLLE = −12λijk
[˜∗Rkνi`j + νi`j`
ck + ˜Lj`ckνi − (i↔ j)
]+ c.c. , (J.2)
LLQD = −λ′ijk(d∗Rkνidj + νidjd
ck + dLjd
ckνi − d∗Rk`iuj − uLjdck`i − ˜Liujdck
)+ c.c. , (J.3)
LUDD = −12λ
′′ijkεc1c2c3
[uc1∗Ri d
cc2j dcc3k + dc2∗Rj u
cc1i dcc3k + dc3∗Rku
cc1i dcc2j
]+ c.c. , (J.4)
where repeated indices are summed over. The extra factors of 12 are convenient due to the
anti-symmetry in the corresponding couplings: λijk = −λjik, λ′′ijk = −λ′′ikj. Using eq. (4.9),
and Fig. 8 we can now directly determine the corresponding Feynman rules. These are given
in Figs. 82, 83, and 84. The same Lagrangian for the Yukawa interactions is given in terms of
4-component fermions in [103, 104], for example.
181
˜Rk νi
`j
−iλijkνi
` ck
`j
−iλijk
˜Lj
` ck
νi
−iλijk
Figure 82: Feynman rules for the Yukawa couplings of two-component fermions due to the su-persymmetric, R-parity violating superpotential terms LLE. For each diagram, there is anotherwith all arrows reversed and λijk → λ∗ijk.
dRk νi
dj
−iλ′ijkνi
dck
dj
−iλ′ijk
dLjdck
νi
−iλ′ijkdRk
`i
uj
iλ′ijk
uLjdck
`i
iλ′ijk ˜Li
dck
uj
iλ′ijk
Figure 83: Feynman rules for the Yukawa couplings of two-component fermions for the super-symmetric, R-parity violating superpotential term LQD. For each diagram, there is anotherwith all arrows reversed and λ′ijk → λ′∗ijk.
uc1Ridcc2j
dcc3k
−iεc1c2c3λ′′ijkdc3Rk
ucc1i
dcc2j
−iεc1c2c3λ′′ijk
Figure 84: Feynman rules for the Yukawa couplings of two-component fermions due to thesupersymmetric, R-parity violating superpotential terms UDD. For each diagram, there isanother with all arrows reversed and λ′′ijk → λ′′∗ijk.
182
References
[1] E. Cartan, Bull. Math. France 41 (1913) 53.
[2] E. Cartan, Theory of Spinors, (Hermann, Paris, 1966). The first edition of this book ap-
peared in 1937.
[3] P. A. M. Dirac, Proc. Royal Soc. (A) 117 (1928) p.610, P. A. M. Dirac, Proc. Royal Soc.
(A) 118 (1928) p.351.
[4] H. Weyl, Gruppentheorie und Quantenmechanik, S. Hirzel Verlag, Leipzig, 1928. (English
translation: The theory of groups and quantum mechanics, translated from the 2d. rev.
German ed. by H. P. Robertson, Dover Publications, New York, 1949.)
[5] B. L. van der Waerden, Nachrichten Akad. Wiss. Gottingen, Math.-physik. Kl., p. 100
(1929)
[6] O. LaPorte, and G. E. Uhlenbeck, Phys. Rev. 37 (1931) 1380.
[7] L. Infeld and B. L. van der Waerden, Sitzber. preuss. Akad. d. Wiss., Physik.-math. Kl.
380 (1933)
[8] B. L. van der Waerden, Group Theory and Quantum Mechanics, Springer-Verlag, Heidel-
berg, 1974. The original version of this book was published in German in 1932.
[9] J. A. Schouten, Journal of Math. and Phys. 10 (1931) p.239.
[10] W. L. Bade and H. Jehle, Rev. Mod. Phys. 25 (1953) 714-728.
[11] F. Cap, Fortschritte der Physik 2, (1954) p.207-231.
[12] E.M. Corson, Introduction to Tensors, Spinors and Relativistic Wave Equations (Blackie
and Son, London, 1953).
[13] H. Umezawa, Quantum Field Theory (North-Holland Publishing Company, Amsterdam,
1956).
[14] P. Roman, Theory of Elementary Particles (North-Holland Publishing Company, Amster-
dam, 1960).
[15] J. Rzewuski, Field Theory Volume 1: Classical Theory (PWN-Polish Scientific Publishers,
Warsaw, 1958, published in Great Britain by Iliffe Books Ltd., London, 1967).
[16] V.B. Berestetskii, E.M. Lifshitz and L.P. Pitaevskii, Relativistic Quantum Theory (Pergam-
mon Press, Oxford, UK, 1971).
183
[17] Peter A. Carruthers, J. Math. Phys., 9 (1968) 1835; Spin and Isospin in Particle Physics
(Gordon and Breach Science Publishers, New York, 1971).
[18] Yu. V. Novozhilov, Introduction to Elementary Particle Theory (Pergammon Press, Oxford,
UK, 1975).
[19] I. Bialynicki-Birula and Z. Bialynicki-Birula, Quantum Electrodynamics (Polish Scienteific
Publishers, Warsaw and Pergammon Press, Oxford, UK, 1975). In this book, the dot-
ted/undotted indices make a brief appearence in Appendix I.
[20] V.B. Berestetskii, E.M. Lifshitz and L.P. Pitaevskii, Quantum Electrodyanmics (Pergam-
mon Press, Oxford, UK, 1980).
[21] F. Scheck, Electromagnetism and Strong Interactions: An Introduction to Theoretical Par-
ticle Physics (2nd edition) (Springer-Verlag, Berlin, 1983).
[22] R. Penrose and W. Rindler, Spinors and Space-Time, volume 1: Two-Spinor Calculus and
Relativistic Fields (Cambridge University Press, Cambridge, UK, 1984).
[23] Roman U. Sexl and Helmuth K. Urbantke, Relativity, Groups, Particles: Special Relativity
and Relativistic Symmetry in Field and Particle Physics (springer-Verlag, Vienna, 1992).
[24] R. Ticciati, Quantum Field Theory for Mathematicians (Cambridge University Press, Cam-
bridge, UK, 1999).
[25] W. Siegel, “Fields,” arXiv:hep-th/9912205.
[26] A.I. Akhiezer and S.V. Peletminsky, Fields and Fundamental Interactions (Taylor & Francis,
London, 2002).
[27] Milutin Blagojevic, Gravitation and Gauge Symmetries (Institute of Physics Publishing,
Bristol, UK, 2002).
[28] Heinrich Saller, Operational Quantum Theory II: Relativistic Structures (Springer-Verlag,
New York, 2006).
[29] Moshe Carmeli and Shimon Malin, Theory of Spinors: An Introduction (World Scientific,
Singapore, 2000).
[30] Jean Hladik, Spinors in Physics (Springer-Verlag, New York, 1999).
[31] Mark Srednicki, Quantum Field Theory (Cambridge University Press, Cambridge, UK,
2007)
184
[32] J. Wess and J. Bagger, Supersymmetry and Supergravity (Princeton University Press,
Princeton, NJ, 1992).
[33] S.J. Gates Jr., M.T. Grisaru, M. Rocek and W. Siegel, Superspace or One Thousand and
One Lessons in Supersymmetry (The Benjamin Cummins Publishing Company, Reading,
MA, 1983).
[34] G. Ross, Grand Unified Theories (Benjamin-Cummings Publishing Co., 1984)
[35] Prem P. Srivastava, Supersymmetry, Superfields and Supergravity: an introduction (Adam
Hilger, Bristol, UK, 1986).
[36] P. West, Introduction to Supersymmetry and Supergravity (World Scientific, Singapore,
1986)
[37] H.J.W. Muller-Kirsten and A. Wiedemann, Supersymmetry: An Introduction with Con-
ceptual and Calculational Details (World Scientific, Singapore, 1987).
[38] D. Bailin and A. Love, Supersymmetric Gauge Fied Theory and String Theory (Institute
of Physics Publishing, Bristol, UK, 1994).
[39] Ioseph L. Buchbinder and Sergei M. Kuzenko, Ideas and Methods of Supersymmetry and
Supergravity or a Wlkf through Superspace (Institute of Physics Publishing, Bristol, UK,
1995).
[40] S.K. Soni and S. Singh, Supersymmetry: Basics and Concepts (Narosa Publishing House,
New Delhi, 2000).
[41] A.S. Galperin, E.A. Ivanov, V.I. Ogievetsky and E.S. Sokatchev, Harmonic Superspace
(Cambridge University Press, Cambridge, UK, 2001).
[42] Manuel. Drees, Rohini M. Godbole and Probir Roy, Theory and Phenomenology of Sparti-
cles (World Scientific, Singapore, 2004).
[43] J. Terning, Modern Supersymmetry (Oxford Science Publications, Oxford, UK, 2006).
[44] Pierre Binetruy, Supersymmetry: Theory, Experiment and Cosmology (Oxford University
Press, Oxford, UK, 2006).
[45] I. J. R. Aitchison, Supersymmetry in Particle Physics: An Elementary Introduction, (Cam-
bridge University Press, Cambridge, UK,2007). An earlier version of this book is available
at arXiv:hep-ph/0505105.
[46] K. M. Case, Phys. Rev. 107 (1957) p.307.
185
[47] H. P. Nilles, Phys. Rept. 110 (1984) 1.
[48] H. E. Haber and G. L. Kane, Phys. Rept. 117 (1985) 75.
[49] S.P. Martin, “A Supersymmetry Primer,” [hep-ph/9709356].
[50] W. Pauli, Annalen Phys. 18S5 (1933) 305; W. Pauli, Annalen Phys. 18S5 (1933) 337;
V. Bargmann, Sitzber. preuss. Akad. Wiss., Physik.-math. Kl., 346 (1932); W. Kofink,
Math. Z. 51 (1949) 702.
[51] J. D. Bjorken and M. C. Chen, Phys. Rev. 154 (1966) 1335.
[52] G. R. Henry, Phys. Rev. 154 (1967) 1534.
[53] H. W. Fearing and R. R. Silbar, Phys. Rev. D 6 (1972) 471.
[54] J. O. Eeg, J. Math. Phys. 21 (1980) 170.
[55] G. Passarino, Phys. Rev. D 28 (1983) 2867.
[56] G. Passarino, Nucl. Phys. B 237 (1984) 249.
[57] R. Kleiss, Nucl. Phys. B 241 (1984) 61.
[58] R. Kleiss and W. J. Stirling, Nucl. Phys. B 262 (1985) 235.
[59] A. Ballestrero and E. Maina, Phys. Lett. B 350 (1995) 225 [arXiv:hep-ph/9403244].
[60] G. Chalmers and W. Siegel, Phys. Rev. D 59 (1999) 045012 [arXiv:hep-ph/9708251];
G. Chalmers and W. Siegel, Phys. Rev. D 59 (1999) 045013 [arXiv:hep-ph/9801220].
[61] G. R. Farrar and F. Neri, Phys. Lett. 130B (1983) 109 [Addendum-ibid. 152B (1985) 443].
[62] A. Kersch and F. Scheck, Nucl. Phys. B 263 (1986) 475.
[63] K. Hagiwara and D. Zeppenfeld, Nucl. Phys. B 274 (1986) 1.
[64] F. A. Berends and W. Giele, Nucl. Phys. B 294 (1987) 700; . A. Berends, W. T. Giele and
H. Kuijf, Nucl. Phys. B 321 (1989) 39.
[65] S. Dittmaier, Phys. Rev. D 59 (1999) 016007 [arXiv:hep-ph/9805445].
[66] S. Dittmaier, “Gauge boson production in electron - photon collisions”, Ph. D. thesis, RX-
1526 (WURZBURG.
[67] J. Schechter and J. W. F. Valle, Phys. Rev. D 22 (1980) 2227.
186
[68] J. Schechter and J. W. F. Valle, Phys. Rev. D 24 (1981) 1883 [Erratum-ibid. D 25 (1982)
283].
[69] T. Takagi, Japan J. Math. 1 (1925) 83.
[70] R.A. Horn and C.R. Johnson, Matrix Analysis (Cambridge University Press, Cambridge,
England, 1990).
[71] D.S. Bernstein, Matrix Mathematics: Theory, Facts, And Formulas With Application To
Linear Systems Theory (Princeton University Press, Princeton, NJ, 2005).
[72] H. J. Bhabha, Proc. Roy. Soc (London) A 154 (1935) 195.
[73] S. Weinberg, Phys. Rev. Lett. 19 (1967) 1264.
[74] M. Kobayashi and T. Maskawa, Prog. Theor. Phys. 49 (1973) 652; N. Cabibbo, Phys. Rev.
Lett. 10 (1963) 531.
[75] V.D. Barger and R.J.N. Phillips, Collider Physics, Addison Wesley Longman, 1996.
[76] M. Peskin and D. Schroeder, Introduction to Quantum Field Theory (Addison Wesley,
Reading, MA, 1995).
[77] L. B. Okun, Leptons and Quarks, North-Holland Publishing Company, Amsterdam, 1982
[78] This was first calculated for electrons in: L. Resnick, M. K. Sundaresan and P. J. S. Watson,
Phys. Rev. D 8 (1973) 172.
[79] J.F. Gunion, H.E. Haber, G. Kane and S. Dawson, The Higgs Hunter’s Guide (Perseus
Publishing, Cambridge, MA, 2000).
[80] D. A. Dicus, S. Nandi and X. Tata, Phys. Lett. B 129 (1983) 451 [Erratum-ibid. B 145
(1984) 448].
[81] J. F. Gunion and H. E. Haber, Phys. Rev. D 37 (1988) 2515.
[82] J. F. Gunion and H. E. Haber, Nucl. Phys. B 272 (1986) 1 [Erratum-ibid. B 402 (1993)
567].
[83] W. Y. Keung and L. Littenberg, Phys. Rev. D 28 (1983) 1067.
[84] R. M. Barnett, K. S. Lackner and H. E. Haber, Phys. Rev. Lett. 51 (1983) 176; Phys. Rev.
D 29 (1984) 1381.
[85] A. Bartl, H. Fraas and W. Majerotto, Nucl. Phys. B 278, 1 (1986).
187
[86] J. R. Ellis, J. M. Frere, J. S. Hagelin, G. L. Kane and S. T. Petcov, Phys. Lett. B 132
(1983) 436; E. Reya, Phys. Lett. B 133 (1983) 245; P. Chiappetta, J. Soffer, P. Taxil,
F. M. Renard and P. Sorba, Nucl. Phys. B 262 (1985) 495 [Erratum-ibid. B 279 (1987)
824];
[87] S. Dawson, E. Eichten and C. Quigg, Phys. Rev. D 31 (1985) 1581.
[88] A. Bartl, H. Fraas and W. Majerotto, Z. Phys. C 30 (1986) 441.
[89] D. A. Dicus, S. Nandi, W. W. Repko and X. Tata, Phys. Rev. Lett. 51 (1983) 1030 [Erratum-
ibid. 51 (1983) 1813]; P. Chiappetta, J. Soffer, P. Taxil, F. M. Renard and P. Sorba, Nucl.
Phys. B 262 (1985) 495 [Erratum-ibid. B 279 (1987) 824].
[90] S. Y. Choi, A. Djouadi, H. K. Dreiner, J. Kalinowski and P. M. Zerwas, Eur. Phys. J. C 7
(1999) 123 [arXiv:hep-ph/9806279].
[91] W. Beenakker, M. Klasen, M. Kramer, T. Plehn, M. Spira and P. M. Zerwas, Phys. Rev.
Lett. 83 (1999) 3780 [arXiv:hep-ph/9906298].
[92] H. Baer, K. Hagiwara and X. Tata, Phys. Rev. D 35 (1987) 1598.
[93] S. Dimopoulos, M. Dine, S. Raby and S. D. Thomas, Phys. Rev. Lett. 76 (1996) 3494
[arXiv:hep-ph/9601367].
[94] G. F. Giudice and R. Rattazzi, Phys. Rept. 322 (1999) 419 [arXiv:hep-ph/9801271].
[95] S. Ambrosanio, G. D. Kribs and S. P. Martin, Nucl. Phys. B 516, 55 (1998) [hep-
ph/9710217].
[96] P. Fayet, Phys. Lett. B 70, 461 (1977); Phys. Lett. B 86, 272 (1979). R. Casalbuoni et al.,
Phys. Lett. B 215, 313 (1988); Phys. Rev. D 39, 2281 (1989).
[97] N. Cabibbo, G.R. Farrar and L. Maiani, Phys. Lett. B 105, 155 (1981), D.R. Stump,
M. Wiest and C.P. Yuan, Phys. Rev. D 54, 1936 (1996) [hep-ph/9601362], S. Dimopoulos,
M. Dine, S. Raby and S.D. Thomas, Phys. Rev. Lett. 76, 3494 (1996) [hep-ph/9601367],
S. Ambrosanio et al., Phys. Rev. Lett. 76, 3498 (1996) [hep-ph/9602239], Phys. Rev. D 54,
5395 (1996) [hep-ph/9605398].
[98] P. R. Harrison and C. H. Llewellyn Smith, Nucl. Phys. B 213, 223 (1983) [Erratum-ibid.
B 223, 542 (1983)],
[99] B. C. Allanach, A. Dedes and H. K. Dreiner, Phys. Rev. D 69 (2004) 115002 [hep-
ph/0309196].
188
[100] B. C. Allanach, M. A. Bernhardt, H. K. Dreiner, C. H. Kom and P. Richardson, arXiv:hep-
ph/0609263.
[101] S. Dimopoulos and L. J. Hall, Phys. Lett. B 207 (1988) 210; J. Erler, J. L. Feng and
N. Polonsky, Phys. Rev. Lett. 78 (1997) 3063 [arXiv:hep-ph/9612397]; H. K. Dreiner,
P. Richardson and M. H. Seymour, Phys. Rev. D 63 (2001) 055008 [arXiv:hep-ph/0007228];
H. K. Dreiner, S. Grab, M. Kramer and M. K. Trenkel, arXiv:hep-ph/0611195.
[102] S. Dimopoulos, R. Esmailzadeh, L. J. Hall and G. D. Starkman, Phys. Rev. D 41 (1990)
2099.
[103] H. K. Dreiner, P. Richardson and M. H. Seymour, JHEP 0004 (2000) 008 [arXiv:hep-
ph/9912407].
[104] P. Richardson, [hep-ph/0101105].
[105] H. K. Dreiner and G. G. Ross, Nucl. Phys. B 365 (1991) 597.
[106] S. Dawson, Nucl. Phys. B 261 (1985) 297.
[107] E. A. Baltz and P. Gondolo, Phys. Rev. D 57 (1998) 2969 [arXiv:hep-ph/9709445].
[108] J. Butterworth and H. K. Dreiner, Nucl. Phys. B 397 (1993) 3 [arXiv:hep-ph/9211204];
H. K. Dreiner, M. Guchait and D. P. Roy, Phys. Rev. D 49 (1994) 3270 [arXiv:hep-
ph/9310291]; H. K. Dreiner and P. Morawitz, Nucl. Phys. B 428, 31 (1994) [Erratum-ibid.
B 574, 874 (2000)] [arXiv:hep-ph/9405253].
[109] Y. Nambu and G. Jona-Lasinio, Phys. Rev. 122, 345 (1961); Phys. Rev. 124, 246 (1961).
[110] Y. Nambu, EFI-88-62-CHICAGO, Proc. of 1988 Int. Workshop New Trends in Strong
Coupling Gauge Theories, Nagoya, Japan, Aug 24-27, 1988; V.A. Miransky, M. Tanabashi
and K. Yamawaki, Mod. Phys. Lett. A 4, 1043 (1989); Phys. Lett. B 221, 177 (1989);
W.J. Marciano, Phys. Rev. Lett. 62, 2793 (1989); Phys. Rev. D 41, 219 (1990);
[111] W.A. Bardeen, C.T. Hill and M. Lindner, Phys. Rev. D 41, 1647 (1990).
[112] D.E. Clague and G.G. Ross, Nucl. Phys. B 364 (1991) 43.
[113] C.T. Hill, Phys. Lett. B 266, 419 (1991).
[114] C.T. Hill and E.H. Simmons, Phys. Rept. 381, 235 (2003) [Erratum-ibid. 390, 553 (2004)]
[hep-ph/0203079].
[115] P. Ramond, Field theory: A modern primer (Perseus Books, 2nd ed. 1994).
189
[116] G. ’t Hooft and M. J. G. Veltman, Nucl. Phys. B 44, 189 (1972). W.A. Bardeen, A.J. Buras,
D.W. Duke and T. Muta, Phys. Rev. D 18, 3998 (1978).
[117] W. Siegel, Phys. Lett. B 84, 193 (1979); D. M. Capper, D. R. T. Jones and P. van
Nieuwenhuizen, Nucl. Phys. B 167, 479 (1980).
[118] R. Hempfling and B.A. Kniehl, Phys. Rev. D 51, 1386 (1995) [hep-ph/9408313].
[119] R. Tarrach, Nucl. Phys. B 183, 384 (1981).
[120] S.P. Martin and M.T. Vaughn, Phys. Lett. B 318, 331 (1993) [hep-ph/9308222].
[121] D.M. Pierce and A. Papadopoulos, Nucl. Phys. B 430, 278 (1994) [hep-ph/9403240],
D.M. Pierce, J.A. Bagger, K.T. Matchev and R.j. Zhang, Nucl. Phys. B 491, 3 (1997)
[hep-ph/9606211].
[122] S. Weinberg, “The Quantum Theory of Fields. Vol. 2: Modern Applications,”
[123] R. Jackiw, in Current Algebra and Anomalies, by S.B. Trieman, R. Jackiw, B. Zumino
and E. Witten (Princeton University Press, Princeton, NJ, 1985).
[124] W. Siegel, Phys. Lett. B 94, 37 (1980).
[125] L. V. Avdeev and A. A. Vladimirov, Nucl. Phys. B 219, 262 (1983).
[126] See for example M. S. Chanowitz, M. Furman and I. Hinchliffe, Nucl. Phys. B 159, 225
(1979), and references therein.
[127] M. Jacob and G.C. Wick, Ann. Phys. (NY) 7 (1959) 404.
[128] R.A. Horn and C.R. Johnson, Topics in Matrix Analysis (Cambridge University Press,
Cambridge, England, 1991).
[129] L. Autonne, Sur les matrices hypohermitiennes et sur les matrices unitaire, Annales de
l’Universite de Lyon, Nouvelle Serie I, Fasc. 38 (1915) 1–77.
[130] S.Y. Choi and M. Drees, unpublished. This proof was inspired by the diagonalization
algorithm of hermitian matrices in W.H. Williams, B.P. Flannery, S.A. Teukolsky and
W.T. Vetterling, Numerical Recipes in Fortran 77 (Cambridge University Press, Cambridge,
England, 1999), section 11.4. A similar method of proof is outlined in Ref. [70], section 4.4,
problem 2 (on pp. 212–213) and section 4.6, problem 15 (on p. 254).
[131] S.Y. Choi, H.E. Haber, J. Kalinowski and P.M. Zerwas, ****
[132] T. Hahn, arXiv:physics/0607103.
190
[133] T. Hahn, Nucl. Phys. Proc. Suppl. 116 (2003) 363 [arXiv:hep-ph/0210220].
[134] T. Hahn, In the Proceedings of 2005 International Linear Collider Workshop (LCWS
2005), Stanford, California, 18-22 Mar 2005, pp 0604 [arXiv:hep-ph/0506201].
[135] C. Bouchiat and L. Michel, Nucl. Phys. 5 (1958) 416; L. Michel, Suppl. Nuovo Cim. 14
(1959) 95.
[136] H.E. Haber, “Spin Formalism and Applications to New Physics Searches,” in the Pro-
ceedings of the 21st SLAC Summer Institute on Particle Physics: Spin Structure in High
Energy Processes, SLAC, Stanford, CA 26 July–6 August 1993, pp. 231–272. [arXiv:hep-
ph/9405376].
[137] R. Vega and J. Wudka, Phys. Rev. D53 (1996) 5286 [Erratum: D56 (1997) 6037],
[arXiv:hep-ph/9511318]
[138] A. Bunse–Gerstner and W.B. Gragg, J. Comp. Appl. Math. 21 (1988) 41; W. Xu and
S. Qiao, “A Divide–and–Conquer Method for the Takagi Factorization,” Technical Report
No. CAS 05–01–SQ, (February 2005).
[139] X. Wang and S. Qiao, in the Proceedings of the International Conference on Parallel and
Distributed Processing Techniques and Applications, Vol. I, edited by H.R. Arabnia, pp.
206–212, Las Vegas, Nevada, USA, June 2002, pp. 206–212; F.T. Luk and S. Qiao, in
Advanced Signal Processing Algorithms, Architectures, and Implementations XI, edited by
F.T. Luk, Proc. SPIE 4474 (2001) 254.
[140] J.M. Jauch and F. Rohrlich, The Theory of Photons and Electrons, second expanded
edition (Springer-Verlag, New York, 1976), Appendix A2-2;
[141] D. Bailin, Weak Interactions, second edition (Adam Hilger Ltd., Bristol, England, 1982),
chapter 2.
[142] K.I. Aoki, Z. Hioki, M. Konuma, R. Kawabe and T. Muta, Prog. Theor. Phys. Suppl. 73
(1982) 1; M. Bohm, H. Spiesberger and W. Hollik, Fortsch. Phys. 34 (1986) 687.
[143] M. Bohm, A. Denner and H. Joos, Gauge theories of the strong and electroweak interaction,
(Teubner, Stuttgart, Germany, 2001).
[144] S. Eidelman et al. [Particle Data Group Collaboration], “Review of particle physics,” Phys.
Lett. B 592 (2004) 1.
[145] H.K. Dreiner, [hep-ph/9707435]; R. Barbier et al., [hep-ph/0406039].
191
[146] S. Weinberg, Phys. Rev. D 26 (1982) 287; N. Sakai and T. Yanagida, Nucl. Phys. B 197
(1982) 533.
[147] A. Dedes, S. Rimmer, J. Rosiek and M. Schmidt-Sommerfeld, Phys. Lett. B 627, 161
(2005) [arXiv:hep-ph/0506209]; A. Dedes, S. Rimmer and J. Rosiek, JHEP 0608, 005
(2006) [arXiv:hep-ph/0603225].
[148] A. Dedes, S. Rimmer and J. Rosiek, JHEP 0608 (2006) 005. [arXiv:hep-ph/0603225].
[149] J.F. Donoghue, Phys. Rev. D19 (1979) 2772.
[150] M. Capdequi Peyranere and M. Talon, Nuovo Cim. A75 (1983) 205.
[151] A. Denner and T. Sack, Nucl. Phys. B347 (1990) 203. A. Denner, Fortsch. Phys. 41 (1993)
307.
[152] B.A. Kniehl and A. Pilaftsis, Nucl. Phys. B474, 286 (1996) [arXiv:hep-ph/9601390]. A. Pi-
laftsis, Phys. Rev. D65 (2002) 115013 [arXiv:hep-ph/0203210].
[153] S. Kiyoura, M. M. Nojiri, D.M. Pierce and Y. Yamada, Phys. Rev. D58 (1998) 075002
[arXiv:hep-ph/9803210]; D.M. Pierce, in Supersymmetry, Supergravity and Supercolliders,
Proceedings of the Theoretical Advanced Study Institute in Elementary Particle Physics
(TASI 97), Bounler, CO, 1–27 Jun 1997, edited by J.A. Bagger (World Scientific, Singapore,
1999) pp. 343–389 [arXiv:hep-ph/9805497].
[154] P. Gambino and P.A. Grassi, Phys. Rev. D62 (2000) 076002 [arXiv:hep-ph/9907254].
[155] A.O. Bouzas, Eur. Phys. J. C20 (2001) 239 [Erratum: ibid. C27 (2003) 623] [arXiv:hep-
ph/0101101].
[156] Y. Yamada, Phys. Rev. D64 (2001) 036008 [arXiv:hep-ph/0103046].
[157] D. Espriu, J. Manzano and P. Talavera, Phys. Rev. D66 (2002) 076002 [arXiv:hep-
ph/0204085].
[158] Y. Zhou, J. Phys. G29 (2003) 1031 [arXiv:hep-ph/0301090]; Mod. Phys. Lett. A21 (2006)
2763 [arXiv:hep-ph/0502186].
[159] Y. Liao, Phys. Rev. D69 (2004) 016001 [arXiv:hep-ph/0309034].
192
Recommended